# gjalla model cards

> What AI models are reported to be good and bad at, with the source and date for every claim. Last updated 2026-10-26.

## Start here
- [Model index](https://gjalla.io/models/): every model, with vendor, tier, price and benchmark percentiles
- [Methodology](https://gjalla.io/models/methodology/): where claims come from, what the source labels mean, how percentiles are computed
- [models.json](https://gjalla.io/models/models.json): all of it as JSON

## Models
- [claude-fable-5](https://gjalla.io/models/claude-fable-5/): anthropic, 2 benchmark results, 0 claims, 0 practitioner reports
- [claude-fable-5.1](https://gjalla.io/models/claude-fable-5.1/): anthropic, 12 benchmark results, 12 claims, 16 practitioner reports
- [claude-haiku-5.5](https://gjalla.io/models/claude-haiku-5.5/): anthropic, 7 benchmark results, 13 claims, 10 practitioner reports
- [claude-mythos-5](https://gjalla.io/models/claude-mythos-5/): anthropic, 1 benchmark results, 0 claims, 0 practitioner reports
- [claude-mythos-5.1](https://gjalla.io/models/claude-mythos-5.1/): anthropic, 1 benchmark results, 0 claims, 2 practitioner reports
- [claude-opus-4.5](https://gjalla.io/models/claude-opus-4.5/): anthropic, 1 benchmark results, 0 claims, 0 practitioner reports
- [claude-opus-4.6](https://gjalla.io/models/claude-opus-4.6/): anthropic, 1 benchmark results, 0 claims, 0 practitioner reports
- [claude-opus-5](https://gjalla.io/models/claude-opus-5/): anthropic, 6 benchmark results, 1 claims, 3 practitioner reports
- [claude-opus-5.5](https://gjalla.io/models/claude-opus-5.5/): anthropic, 10 benchmark results, 17 claims, 22 practitioner reports
- [claude-sonnet-4.5](https://gjalla.io/models/claude-sonnet-4.5/): anthropic, 1 benchmark results, 0 claims, 0 practitioner reports
- [claude-sonnet-5](https://gjalla.io/models/claude-sonnet-5/): anthropic, 6 benchmark results, 4 claims, 3 practitioner reports
- [claude-sonnet-5.5](https://gjalla.io/models/claude-sonnet-5.5/): anthropic, 12 benchmark results, 16 claims, 16 practitioner reports
- [deepseek-v3.2](https://gjalla.io/models/deepseek-v3.2/): deepseek, 1 benchmark results, 0 claims, 0 practitioner reports
- [deepseek-v4-flash-vision-exp](https://gjalla.io/models/deepseek-v4-flash-vision-exp/): deepseek, 1 benchmark results, 0 claims, 0 practitioner reports
- [deepseek-v4-pro](https://gjalla.io/models/deepseek-v4-pro/): deepseek, 5 benchmark results, 5 claims, 4 practitioner reports
- [deepseek-v4.1-flash](https://gjalla.io/models/deepseek-v4.1-flash/): deepseek, 8 benchmark results, 14 claims, 8 practitioner reports
- [devstral-2](https://gjalla.io/models/devstral-2/): mistral, 1 benchmark results, 1 claims, 0 practitioner reports
- [gemini-3-flash](https://gjalla.io/models/gemini-3-flash/): google, 7 benchmark results, 0 claims, 1 practitioner reports
- [gemini-3-pro](https://gjalla.io/models/gemini-3-pro/): google, 1 benchmark results, 0 claims, 0 practitioner reports
- [gemini-3.1-pro](https://gjalla.io/models/gemini-3.1-pro/): google, 6 benchmark results, 9 claims, 8 practitioner reports
- [gemini-3.5-flash-lite](https://gjalla.io/models/gemini-3.5-flash-lite/): google, 6 benchmark results, 9 claims, 4 practitioner reports
- [gemini-3.7-flash](https://gjalla.io/models/gemini-3.7-flash/): google, 2 benchmark results, 0 claims, 0 practitioner reports
- [gemini-3.8-flash](https://gjalla.io/models/gemini-3.8-flash/): google, 6 benchmark results, 12 claims, 15 practitioner reports
- [gemini-4-argon](https://gjalla.io/models/gemini-4-argon/): google, 6 benchmark results, 14 claims, 13 practitioner reports
- [glm-4.6](https://gjalla.io/models/glm-4.6/): zhipu, 1 benchmark results, 0 claims, 0 practitioner reports
- [glm-5.2](https://gjalla.io/models/glm-5.2/): zhipu, 6 benchmark results, 8 claims, 12 practitioner reports
- [glm-5.3](https://gjalla.io/models/glm-5.3/): zhipu, 6 benchmark results, 4 claims, 2 practitioner reports
- [glm-5.3-flash](https://gjalla.io/models/glm-5.3-flash/): zhipu, 1 benchmark results, 0 claims, 0 practitioner reports
- [gpt-5](https://gjalla.io/models/gpt-5/): openai, 1 benchmark results, 0 claims, 0 practitioner reports
- [gpt-5.2](https://gjalla.io/models/gpt-5.2/): openai, 1 benchmark results, 0 claims, 0 practitioner reports
- [gpt-5.2-codex](https://gjalla.io/models/gpt-5.2-codex/): openai, 1 benchmark results, 0 claims, 0 practitioner reports
- [gpt-5.3-codex](https://gjalla.io/models/gpt-5.3-codex/): openai, 2 benchmark results, 0 claims, 1 practitioner reports
- [gpt-5.4](https://gjalla.io/models/gpt-5.4/): openai, 1 benchmark results, 0 claims, 0 practitioner reports
- [gpt-5.5](https://gjalla.io/models/gpt-5.5/): openai, 2 benchmark results, 2 claims, 4 practitioner reports
- [gpt-5.6-cyber](https://gjalla.io/models/gpt-5.6-cyber/): openai, 0 benchmark results, 0 claims, 4 practitioner reports
- [gpt-5.6-luna](https://gjalla.io/models/gpt-5.6-luna/): openai, 2 benchmark results, 0 claims, 4 practitioner reports
- [gpt-5.6-sol](https://gjalla.io/models/gpt-5.6-sol/): openai, 6 benchmark results, 9 claims, 12 practitioner reports
- [gpt-5.6-terra](https://gjalla.io/models/gpt-5.6-terra/): openai, 2 benchmark results, 0 claims, 6 practitioner reports
- [gpt-6-astra](https://gjalla.io/models/gpt-6-astra/): openai, 9 benchmark results, 14 claims, 14 practitioner reports
- [gpt-6-luna](https://gjalla.io/models/gpt-6-luna/): openai, 6 benchmark results, 11 claims, 7 practitioner reports
- [gpt-6-sol](https://gjalla.io/models/gpt-6-sol/): openai, 9 benchmark results, 4 claims, 7 practitioner reports
- [gpt-6.1-sol](https://gjalla.io/models/gpt-6.1-sol/): openai, 8 benchmark results, 16 claims, 15 practitioner reports
- [grok-4.5](https://gjalla.io/models/grok-4.5/): xai, 1 benchmark results, 0 claims, 0 practitioner reports
- [grok-4.7](https://gjalla.io/models/grok-4.7/): xai, 7 benchmark results, 5 claims, 1 practitioner reports
- [hy4-preview](https://gjalla.io/models/hy4-preview/): tencent, 2 benchmark results, 0 claims, 0 practitioner reports
- [kimi-k2](https://gjalla.io/models/kimi-k2/): moonshot, 1 benchmark results, 0 claims, 0 practitioner reports
- [kimi-k3](https://gjalla.io/models/kimi-k3/): moonshot, 7 benchmark results, 10 claims, 10 practitioner reports
- [kimi-k3.1-pro](https://gjalla.io/models/kimi-k3.1-pro/): moonshot, 1 benchmark results, 5 claims, 0 practitioner reports
- [mimo-v2.6-flash](https://gjalla.io/models/mimo-v2.6-flash/): xiaomi, 2 benchmark results, 0 claims, 0 practitioner reports
- [mimo-v2.6-pro](https://gjalla.io/models/mimo-v2.6-pro/): xiaomi, 3 benchmark results, 0 claims, 0 practitioner reports
- [mistral-large-4](https://gjalla.io/models/mistral-large-4/): mistral, 3 benchmark results, 10 claims, 8 practitioner reports
- [muse-spark](https://gjalla.io/models/muse-spark/): meta, 1 benchmark results, 0 claims, 0 practitioner reports
- [muse-spark-1.1](https://gjalla.io/models/muse-spark-1.1/): meta, 1 benchmark results, 0 claims, 0 practitioner reports
- [muse-spark-1.3](https://gjalla.io/models/muse-spark-1.3/): meta, 3 benchmark results, 0 claims, 0 practitioner reports
- [muse-wonder](https://gjalla.io/models/muse-wonder/): meta, 2 benchmark results, 0 claims, 0 practitioner reports
- [muse-wonder-full](https://gjalla.io/models/muse-wonder-full/): meta, 3 benchmark results, 5 claims, 1 practitioner reports
- [qwen3-coder-480b](https://gjalla.io/models/qwen3-coder-480b/): alibaba, 1 benchmark results, 0 claims, 0 practitioner reports
- [qwen3.8-27b](https://gjalla.io/models/qwen3.8-27b/): alibaba, 2 benchmark results, 0 claims, 0 practitioner reports
- [qwen3.8-flash-next](https://gjalla.io/models/qwen3.8-flash-next/): alibaba, 1 benchmark results, 0 claims, 0 practitioner reports
- [qwen3.8-max](https://gjalla.io/models/qwen3.8-max/): alibaba, 9 benchmark results, 11 claims, 9 practitioner reports