Publish UnbindAI-reported results and cybersecurity research positioning

.gitattributes

@@ -48,3 +48,14 @@ assets/benchmarks/07-agents.png filter=lfs diff=lfs merge=lfs -text
 assets/benchmarks/08-apex.png filter=lfs diff=lfs merge=lfs -text
 assets/benchmarks/benchmark-overview.gif filter=lfs diff=lfs merge=lfs -text
 assets/benchmarks/comparison-sheet.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/01-aa-index-v5.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/02-hle-diamond.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/03-gpqa-diamond.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/04-terminal-bench-4.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/05-frontiercode-2.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/06-livebench-2.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/07-livecodebench.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/08-full-comparison.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/09-coding-comparison.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/10-reported-320-tests.png filter=lfs diff=lfs merge=lfs -text
+assets/results-v1/preview-sheet.png filter=lfs diff=lfs merge=lfs -text

BASE_MODEL_REFERENCES.md

@@ -0,0 +1,65 @@
+# Unbind Flash V1 — benchmark comparisons
+
+Unbind Flash V1 launch graphics. The highlighted “Unbind Flash V1 · base model” row represents the unmodified DeepSeek-V4.1-Flash-1017 reference checkpoint, using scores reported by DeepSeek. These are base-model comparisons, not evaluations of Unbind’s endpoint, abliterated hosted checkpoint or older public GGUF.
+
+Each chart compares models within one benchmark on a 0–100 axis. Evaluation setups differ across benchmarks; do not average these scores into an overall rating.
+
+[Source table and evaluation notes](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017).
+
+## Animated overview
+
+![Eight benchmark comparisons](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/benchmark-overview.gif)
+
+## DeepSWE v1.1
+
+![DeepSWE v1.1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/01-deepswe.png)
+
+Mini-SWE-V2 harness. Publisher-reported benchmark scores.
+
+## Terminal-Bench 4.0
+
+![Terminal-Bench 4.0 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/02-terminal.png)
+
+Claude Code harness. Publisher-reported benchmark scores.
+
+## HLE without tools
+
+![HLE without tools comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/03-hle.png)
+
+Full HLE figures only. GLM-5.3 omitted: the source gives a text-only subset score.
+
+## ProgramBench
+
+![ProgramBench comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/04-program.png)
+
+Publisher-reported benchmark scores. Additional per-row settings are not specified.
+
+## NL2Repo-Bench
+
+![NL2Repo-Bench comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/05-nl2repo.png)
+
+Publisher-reported benchmark scores. Additional per-row settings are not specified.
+
+## Automation-Bench v1.0.6
+
+![Automation-Bench v1.0.6 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/06-automation.png)
+
+Publisher-reported benchmark scores. Additional per-row settings are not specified.
+
+## Agents’ Last Exam
+
+![Agents’ Last Exam comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/07-agents.png)
+
+Publisher-reported benchmark scores. Additional per-row settings are not specified.
+
+## Apex Bench v1.1
+
+![Apex Bench v1.1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/08-apex.png)
+
+Publisher-reported benchmark scores. Additional per-row settings are not specified.
+
+## Data and reuse
+
+[CSV](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/source-data.csv) · [JSON](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/source-data.json). The accompanying SVG files provide editable vector exports.
+
+All plotted values reproduce the source table. The highlighted display label is branded as “Unbind Flash V1 · base model”; its exact source model is retained in the footer and data files. Other model labels reproduce the source table. GLM-5.3 is omitted from HLE because only its text-only subset score is supplied. These are reported benchmark comparisons, not newly run evaluations.

BENCHMARKS.md

@@ -1,65 +1,53 @@
-# Unbind Flash V1 — benchmark comparisons
-
-Unbind Flash V1 launch graphics. The highlighted “Unbind Flash V1 · base model” row represents the unmodified DeepSeek-V4.1-Flash-1017 reference checkpoint, using scores reported by DeepSeek. These are base-model comparisons, not evaluations of Unbind’s endpoint, abliterated hosted checkpoint or older public GGUF.
-
-Each chart compares models within one benchmark on a 0–100 axis. Evaluation setups differ across benchmarks; do not average these scores into an overall rating.
-
-[Source table and evaluation notes](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017).
-
-## Animated overview
-
-![Eight benchmark comparisons](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/benchmark-overview.gif)
-
-## DeepSWE v1.1
-
-![DeepSWE v1.1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/01-deepswe.png)
-
-Mini-SWE-V2 harness. Publisher-reported benchmark scores.
-
-## Terminal-Bench 4.0
-
-![Terminal-Bench 4.0 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/02-terminal.png)
-
-Claude Code harness. Publisher-reported benchmark scores.
-
-## HLE without tools
-
-![HLE without tools comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/03-hle.png)
-
-Full HLE figures only. GLM-5.3 omitted: the source gives a text-only subset score.
-
-## ProgramBench
-
-![ProgramBench comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/04-program.png)
-
-Publisher-reported benchmark scores. Additional per-row settings are not specified.
-
-## NL2Repo-Bench
-
-![NL2Repo-Bench comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/05-nl2repo.png)
-
-Publisher-reported benchmark scores. Additional per-row settings are not specified.
-
-## Automation-Bench v1.0.6
-
-![Automation-Bench v1.0.6 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/06-automation.png)
-
-Publisher-reported benchmark scores. Additional per-row settings are not specified.
-
-## Agents’ Last Exam
-
-![Agents’ Last Exam comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/07-agents.png)
-
-Publisher-reported benchmark scores. Additional per-row settings are not specified.
-
-## Apex Bench v1.1
-
-![Apex Bench v1.1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/08-apex.png)
-
-Publisher-reported benchmark scores. Additional per-row settings are not specified.
-
-## Data and reuse
-
-[CSV](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/source-data.csv) · [JSON](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/source-data.json). The accompanying SVG files provide editable vector exports.
-
-All plotted values reproduce the source table. The highlighted display label is branded as “Unbind Flash V1 · base model”; its exact source model is retained in the footer and data files. Other model labels reproduce the source table. GLM-5.3 is omitted from HLE because only its text-only subset score is supplied. These are reported benchmark comparisons, not newly run evaluations.
+# Unbind Flash V1 — reported results gallery
+
+Ten Unbind-branded graphics using figures supplied by UnbindAI. These are reported hosted-version results, not independently verified measurements or tests of this repository’s older GGUF download. Settings, dataset revisions, raw logs, and competitor provenance were not supplied.
+
+* Asterisks are retained exactly as supplied; their meaning was not specified.
+
+## Full comparison
+
+![Seven benchmarks, nine models](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/08-full-comparison.png)
+
+## Coding comparison
+
+![Coding benchmark comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/09-coding-comparison.png)
+
+## AA Index v5
+
+![AA Index v5 — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/01-aa-index-v5.png)
+
+## HLE-Diamond
+
+![HLE-Diamond — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/02-hle-diamond.png)
+
+## GPQA Diamond
+
+![GPQA Diamond — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/03-gpqa-diamond.png)
+
+## Terminal-Bench 4
+
+![Terminal-Bench 4 — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/04-terminal-bench-4.png)
+
+## FrontierCode 2
+
+![FrontierCode 2 — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/05-frontiercode-2.png)
+
+## LiveBench 2
+
+![LiveBench 2 — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/06-livebench-2.png)
+
+## LiveCodeBench
+
+![LiveCodeBench — Unbind Flash V1 comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/07-livecodebench.png)
+
+## Reported 320-test result
+
+![Reported 320/320 tests passed](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/10-reported-320-tests.png)
+
+UnbindAI reports 320/320 tests passed. Suite, pass criteria, and logs were not supplied. This is not a universal non-refusal claim or a measured comparison of competitor refusal rates.
+
+## Data and exports
+
+[CSV comparison table](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/comparison-table.csv) · [Source data and provenance](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/source-data.json) · [Asset manifest](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/manifest.json)
+
+Every graphic has a matching editable SVG under `assets/results-v1/`. Charts are sorted by supplied score; tables preserve the supplied model order. No combined score has been computed.

README.md

@@ -7,61 +7,69 @@ pipeline_tag: text-generation
 tags:
 - gguf
 - quantized
-- deepseek
 - conversational
+- cybersecurity
+- penetration-testing
+- uncensored
 ---
 
 ![Unbind Flash V1 — Go off script.](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/unbind-flash-v1.png)
 
 # Unbind Flash V1
 
-**Go off script.**
+**Uncensored AI for cybersecurity research and authorized penetration testing.**
 
-Today we're introducing Unbind Flash V1: an uncensored hosted edition and a separate public GGUF download.
+Security research starts with difficult questions. Unbind Flash V1 is an uncensored hosted edition for people who want to investigate how systems fail, examine code closely, and work through technical problems in depth.
 
-Unbind is for the question you haven't quite figured out how to ask. The idea that sounds unreasonable. The argument worth taking apart. We want conversations with room to develop—and a product that invites you to keep going.
+Our focus is practical: vulnerability analysis, secure code review, capture-the-flag challenges, adversary simulation, and penetration testing within an authorized scope. Bring the suspicious code, the unexpected behavior, or the hypothesis you want to challenge. Keep asking. Follow the evidence.
 
-We're excited to put Unbind into people's hands. Explore it, challenge it, and tell us where it needs work.
+We're releasing Unbind Flash V1 with a comparison of the results reported by UnbindAI, including where other models score higher. The ambition is a useful research partner: capable, direct, and open to difficult technical conversations. These general reasoning and coding results are a starting point; they do not establish performance on dedicated cybersecurity benchmarks.
 
-## Choose your edition
+**Go off script. Stay rigorous.**
 
-| Edition | What it contains |
-|:--|:--|
-| **Hosted Unbind Flash V1** | A separate hosted experience using the [dealignai 1017 UNCENSORED FP8 checkpoint](https://huggingface.co/dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8), identified by UnbindAI. |
-| **This repository's download** | The unchanged [ggml-org DeepSeek-V4-Flash Q2_K_S GGUF](https://huggingface.co/ggml-org/DeepSeek-V4-Flash-GGUF): two shards, approximately **98.6 GB**, MIT licensed. This is an earlier model and is not the hosted abliterated checkpoint. |
-
-## Published results for the hosted checkpoint
+## Results reported by UnbindAI
 
-These are the source publisher's measurements of the checkpoint identified for hosted Unbind. They are not Unbind endpoint tests or measurements of this repository's GGUF files.
+The following figures were supplied by UnbindAI for its hosted Unbind Flash V1 version. They have not been independently verified and do not evaluate this repository's older GGUF download. Exact test settings, dataset revisions, run logs, and competitor provenance were not supplied with this release.
 
-![Publisher-reported MMLU comparison](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/hosted-capability.png)
+| Benchmark | Unbind Flash V1 |
+|:--|--:|
+| AA Index v5 | 31.0* |
+| HLE-Diamond | 26.9 |
+| GPQA Diamond | 93.1 |
+| Terminal-Bench 4 | 12.6 |
+| FrontierCode 2 | 10.9 |
+| LiveBench 2 | 64.6 |
+| LiveCodeBench | 87.2 |
 
-The publisher reports **83.77% (955/1,140)** versus **87.63% (999/1,140)** for the base on a 57-subject MMLU sample. Prompts and decoding matched; serving stacks differed.
+\* Asterisks are retained from the supplied results. Their meaning was not specified.
 
-![Publisher-reported HarmBench classifications](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/hosted-refusal.png)
+![Unbind Flash V1 compared across seven benchmarks](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/08-full-comparison.png)
 
-The publisher reports **320/320 classified COMPLY**, including **0/320 explicit refusals**, with max reasoning and a 200-new-token cap. Its classifier can score the opening reasoning trace when no final answer is available. This result does not establish completed answers or universal compliance.
+[Explore all seven benchmark charts, the coding comparison, and downloadable data](https://huggingface.co/UnbindAI/Unbind-Flash-V1/blob/main/BENCHMARKS.md).
 
-[Read the source evaluation and method](https://huggingface.co/dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8#internal-benchmarks-greedy-t0-this-repo).
+## Reported 320-test result
 
-## Unbind Flash V1 benchmark gallery
+UnbindAI reports **320 tests passed out of 320** for its version. This is a reported test-run outcome, not evidence that every possible request will be answered. The suite name, pass criteria, and raw logs were not supplied; no competitor refusal results were provided.
 
-Eight launch graphics for Unbind Flash V1. The highlighted “Unbind Flash V1 · base model” row represents the unmodified DeepSeek-V4.1-Flash-1017 reference checkpoint, using publisher-reported scores. These comparisons do not measure the Unbind endpoint or abliterated hosted edition.
+![Unbind Flash V1 — reported 320 out of 320 tests passed](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/results-v1/10-reported-320-tests.png)
 
-![Eight base-model benchmark comparisons](https://huggingface.co/UnbindAI/Unbind-Flash-V1/resolve/main/assets/benchmarks/benchmark-overview.gif)
+## Hosted edition and downloadable files
 
-[Browse all eight charts, methodology notes and source data](https://huggingface.co/UnbindAI/Unbind-Flash-V1/blob/main/BENCHMARKS.md).
+| Edition | What it contains |
+|:--|:--|
+| **Hosted Unbind Flash V1** | UnbindAI's uncensored hosted edition, identified by UnbindAI as using the [dealignai 1017 UNCENSORED FP8 checkpoint](https://huggingface.co/dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8). The newly supplied results above refer to this hosted version. |
+| **This repository's download** | The unchanged [ggml-org DeepSeek-V4-Flash Q2_K_S GGUF](https://huggingface.co/ggml-org/DeepSeek-V4-Flash-GGUF): two shards, approximately **98.6 GB**. This is an earlier model, not the hosted abliterated checkpoint. The new results do not apply to these files. |
 
-## Download details
+Download both GGUF shards from **Files and versions**. Runtime memory requirements exceed the file size and depend on software, hardware, and context length. The source notes that this Q2 quantization does not use imatrix calibration.
 
-Download both GGUF shards from **Files and versions**. Runtime memory requirements exceed the file size and depend on your software, hardware and context length. The source notes that this Q2 quantization does not use imatrix calibration.
+UnbindAI created this distribution's branding and documentation. The downloadable weights were copied unchanged; UnbindAI has not trained or fine-tuned them.
 
-The downloadable weights were copied unchanged. UnbindAI created this distribution's branding and documentation and has not trained, fine-tuned or independently evaluated these weights. The hosted edition uses a different checkpoint.
+## Evaluation assets and lineage
 
-## Credits and license
+The [current benchmark gallery](https://huggingface.co/UnbindAI/Unbind-Flash-V1/blob/main/BENCHMARKS.md) contains the newly supplied UnbindAI results. [Earlier base-model reference comparisons](https://huggingface.co/UnbindAI/Unbind-Flash-V1/blob/main/BASE_MODEL_REFERENCES.md) remain separately labeled as upstream measurements.
 
-The downloadable base model was developed by DeepSeek; the GGUF conversion and quantization were published by ggml-org. The hosted checkpoint's weight-level modification was published by dealignai. Both sources identify their license as MIT; this repository retains the original [MIT license notice](https://huggingface.co/UnbindAI/Unbind-Flash-V1/blob/main/LICENSE).
+The downloadable base model was developed by DeepSeek; its GGUF conversion and quantization were published by ggml-org. The hosted checkpoint's weight-level modification was published by dealignai. Both sources identify their license as MIT; this repository retains the original [MIT license notice](https://huggingface.co/UnbindAI/Unbind-Flash-V1/blob/main/LICENSE).
 
 ---
 
-**Unbind** · Your mind, unbound.
+**Unbind · Your mind, unbound.**

assets/results-v1/README.txt

@@ -0,0 +1,17 @@
+UNBIND FLASH V1 — SUPPLIED RESULTS GRAPHICS
+
+10 graphics, each supplied as PNG and editable SVG:
+01–07: individual benchmark comparisons
+08: complete seven-benchmark comparison table
+09: coding comparison table
+10: reported 320/320 test result
+preview-sheet.png: visual index
+comparison-table.csv: exact supplied table
+source-data.json: exact values, asterisks and claim provenance
+manifest.json: image dimensions and asset inventory
+
+Figures supplied by UnbindAI in chat. No external verification performed.
+All asterisks are retained. Their definition was not supplied.
+Benchmark charts are sorted by supplied score; comparison tables retain supplied model order.
+No average, combined score, unsupported win, competitor refusal claim, or universal uncensored claim was added.
+The separate 320/320 card records the user-reported result; suite, criteria and raw logs were not supplied.

assets/results-v1/comparison-table.csv

@@ -0,0 +1,10 @@
+"Model","AA Index v5","HLE-Diamond","GPQA Diamond","Terminal-Bench 4","FrontierCode 2","LiveBench 2","LiveCodeBench"
+"Claude Opus 5","45.0*","49.0*","93.9","49.0","35.0*","78.0*","89.0"
+"GPT-5.6 Sol","39.5*","40.0*","93.5","39.9","28.0*","73.0*","82.6"
+"GPT-6 Sol","43.0*","45.2","94.3","43.9","32.0*","77.0*","87.0*"
+"Claude Sonnet 5","30.0*","30.0*","80.3","14.1","20.0*","68.0*","82.4"
+"GPT-5.6 Terra","34.0*","28.8","93.3","35.4","25.0*","67.8","85.9"
+"Gemini 3.1 Pro","30.0*","31.4","94.4","4.0","18.0*","69.0*","88.5"
+"Muse Spark 1.3","36.4","33.7","93.5","33.3","23.1","63.0","85.0*"
+"Claude Fable 5","38.0*","45.0*","85.9","42.4","30.0*","76.0*","89.8"
+"Unbind Flash V1","31.0*","26.9","93.1","12.6","10.9","64.6","87.2"

assets/results-v1/manifest.json

@@ -0,0 +1,86 @@
+{
+  "source": "Data supplied by UnbindAI in this chat; no external verification requested or performed.",
+  "asterisk_note": "Asterisks are preserved exactly as supplied. Their meaning was not specified.",
+  "assets": [
+    {
+      "id": "01-aa-index-v5",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "01-aa-index-v5.png",
+      "svg": "01-aa-index-v5.svg"
+    },
+    {
+      "id": "02-hle-diamond",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "02-hle-diamond.png",
+      "svg": "02-hle-diamond.svg"
+    },
+    {
+      "id": "03-gpqa-diamond",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "03-gpqa-diamond.png",
+      "svg": "03-gpqa-diamond.svg"
+    },
+    {
+      "id": "04-terminal-bench-4",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "04-terminal-bench-4.png",
+      "svg": "04-terminal-bench-4.svg"
+    },
+    {
+      "id": "05-frontiercode-2",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "05-frontiercode-2.png",
+      "svg": "05-frontiercode-2.svg"
+    },
+    {
+      "id": "06-livebench-2",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "06-livebench-2.png",
+      "svg": "06-livebench-2.svg"
+    },
+    {
+      "id": "07-livecodebench",
+      "width": 1600,
+      "height": 1200,
+      "kind": "benchmark",
+      "png": "07-livecodebench.png",
+      "svg": "07-livecodebench.svg"
+    },
+    {
+      "id": "08-full-comparison",
+      "width": 2400,
+      "height": 1500,
+      "kind": "table",
+      "png": "08-full-comparison.png",
+      "svg": "08-full-comparison.svg"
+    },
+    {
+      "id": "09-coding-comparison",
+      "width": 2400,
+      "height": 1500,
+      "kind": "table",
+      "png": "09-coding-comparison.png",
+      "svg": "09-coding-comparison.svg"
+    },
+    {
+      "id": "10-reported-320-tests",
+      "width": 1600,
+      "height": 1100,
+      "kind": "reported-test-result",
+      "png": "10-reported-320-tests.png",
+      "svg": "10-reported-320-tests.svg"
+    }
+  ]
+}

assets/results-v1/source-data.json

@@ -0,0 +1,187 @@
+{
+  "title": "Unbind Flash V1 benchmark comparison",
+  "source": "Data supplied by UnbindAI in this chat; no external verification requested or performed.",
+  "asterisk_note": "Asterisks are preserved exactly as supplied. Their meaning was not specified.",
+  "benchmarks": [
+    {
+      "id": "01-aa-index-v5",
+      "name": "AA Index v5",
+      "short": [
+        "AA Index",
+        "v5"
+      ],
+      "kind": "Index points"
+    },
+    {
+      "id": "02-hle-diamond",
+      "name": "HLE-Diamond",
+      "short": [
+        "HLE-",
+        "Diamond"
+      ],
+      "kind": "Reported score"
+    },
+    {
+      "id": "03-gpqa-diamond",
+      "name": "GPQA Diamond",
+      "short": [
+        "GPQA",
+        "Diamond"
+      ],
+      "kind": "Reported score"
+    },
+    {
+      "id": "04-terminal-bench-4",
+      "name": "Terminal-Bench 4",
+      "short": [
+        "Terminal-",
+        "Bench 4"
+      ],
+      "kind": "Reported score"
+    },
+    {
+      "id": "05-frontiercode-2",
+      "name": "FrontierCode 2",
+      "short": [
+        "FrontierCode",
+        "2"
+      ],
+      "kind": "Reported score"
+    },
+    {
+      "id": "06-livebench-2",
+      "name": "LiveBench 2",
+      "short": [
+        "LiveBench",
+        "2"
+      ],
+      "kind": "Reported score"
+    },
+    {
+      "id": "07-livecodebench",
+      "name": "LiveCodeBench",
+      "short": [
+        "LiveCodeBench"
+      ],
+      "kind": "Reported score"
+    }
+  ],
+  "models": [
+    {
+      "name": "Claude Opus 5",
+      "scores": [
+        "45.0*",
+        "49.0*",
+        "93.9",
+        "49.0",
+        "35.0*",
+        "78.0*",
+        "89.0"
+      ]
+    },
+    {
+      "name": "GPT-5.6 Sol",
+      "scores": [
+        "39.5*",
+        "40.0*",
+        "93.5",
+        "39.9",
+        "28.0*",
+        "73.0*",
+        "82.6"
+      ]
+    },
+    {
+      "name": "GPT-6 Sol",
+      "scores": [
+        "43.0*",
+        "45.2",
+        "94.3",
+        "43.9",
+        "32.0*",
+        "77.0*",
+        "87.0*"
+      ]
+    },
+    {
+      "name": "Claude Sonnet 5",
+      "scores": [
+        "30.0*",
+        "30.0*",
+        "80.3",
+        "14.1",
+        "20.0*",
+        "68.0*",
+        "82.4"
+      ]
+    },
+    {
+      "name": "GPT-5.6 Terra",
+      "scores": [
+        "34.0*",
+        "28.8",
+        "93.3",
+        "35.4",
+        "25.0*",
+        "67.8",
+        "85.9"
+      ]
+    },
+    {
+      "name": "Gemini 3.1 Pro",
+      "scores": [
+        "30.0*",
+        "31.4",
+        "94.4",
+        "4.0",
+        "18.0*",
+        "69.0*",
+        "88.5"
+      ]
+    },
+    {
+      "name": "Muse Spark 1.3",
+      "scores": [
+        "36.4",
+        "33.7",
+        "93.5",
+        "33.3",
+        "23.1",
+        "63.0",
+        "85.0*"
+      ]
+    },
+    {
+      "name": "Claude Fable 5",
+      "scores": [
+        "38.0*",
+        "45.0*",
+        "85.9",
+        "42.4",
+        "30.0*",
+        "76.0*",
+        "89.8"
+      ]
+    },
+    {
+      "name": "Unbind Flash V1",
+      "scores": [
+        "31.0*",
+        "26.9",
+        "93.1",
+        "12.6",
+        "10.9",
+        "64.6",
+        "87.2"
+      ]
+    }
+  ],
+  "reported_test_result": {
+    "passed": 320,
+    "total": 320,
+    "source": "Reported by UnbindAI",
+    "suite_name": null,
+    "pass_criteria": null,
+    "raw_logs_supplied": false
+  }
+}