Search for a command to run...
A roster-relative index of coding-agent models across five harnesses, with every figure sourced.
Data snapshot: July 28, 2026
The four capability classes are anchor-relative composite-index tiers, not vendor brand names — a model of any vendor can land in any class:
The five evidence grades describe how each figure was sourced:
Coverage is the weighted share of the composite a model's reported benchmarks cover (SWE-bench Verified 25% + SWE-bench Pro 25% + Terminal-Bench 2.1 20% + GPQA Diamond 30% = 100%), not a simple count of benchmarks reported.
Every figure links to the leaderboard or vendor page it came from. These are the benchmark operators whose figures appear here, with their republication terms.
Showing the lowest available rate across harnesses for each model
| Model | Vendor | Harnesses | Class | SWE-bench Verified | SWE-bench Pro | Terminal-Bench 2.1 | GPQA Diamond | Composite index | Coverage | Input price | Output price |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Claude Fable 5 | Anthropic | Claude Code, Cursor, OpenCode Zen | Sonnet | 95.0%self-reported | 80.3%self-reported | 84.3%self-reported | Not reported | 97.3verified | 70% | $10.00verified | $50.00verified |
| Claude Opus 5 | Anthropic | Claude Code, Cursor, OpenCode Zen | Opus | 96.0%self-reported | Not reported | Not reported | 93.2% to 94.3%conflicted | 99.5verified | 55% | $5.00verified | $25.00verified |
| Claude Opus 4.8 | Anthropic | Claude Code, Cursor, OpenCode Zen | Sonnet | 88.6%verified | 69.2%verified | Not reported | Not reported | 89.2verified | 50% | $5.00verified | $25.00verified |
| Claude Sonnet 5 | Anthropic | Claude Code, Cursor, OpenCode Zen | Sonnet | 85.2%self-reported | 63.2%self-reported | 80.4%self-reported | Not reported | 84.8verified | 70% | $3.00verified | $15.00verified |
| Claude Sonnet 4.6 | Anthropic | Claude Code, Cursor, OpenCode Zen | Light | 79.6%secondary | Not reported | Not reported | 74.1% to 89.9%conflicted | 80.6verified | 55% | $3.00verified | $15.00verified |
| Claude Haiku 4.5 | Anthropic | Claude Code, Cursor, OpenCode Zen | Light | 73.3%verified | 39.5%secondary | Not reported | 67.2% to 74.1%conflicted | 66.0verified | 80% | $1.00verified | $5.00verified |
| GPT-5.6 SolIntegrity note | OpenAI | Codex CLI, Cursor, OpenCode Zen | Opus | Not reported | Not reported | 91.9%self-reported | 94.1%secondary | 100.0verified | 50% | $5.00verified | $30.00verified |
| GPT-5.6 Terra | OpenAI | Codex CLI, Cursor, OpenCode Zen | Sonnet | Not reported | Not reported | 87.4%self-reported | Not reported | 95.1verified | 20%low coverage | $2.50verified | $15.00verified |
| GPT-5.6 Luna | OpenAI | Codex CLI, Cursor, OpenCode Zen | Sonnet | Not reported | Not reported | 84.7%self-reported | Not reported | 92.2verified | 20%low coverage | $1.00verified | $6.00verified |
| GPT-5.5 | OpenAI | Codex CLI, Cursor, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $5.00verified | $30.00verified |
| GPT-5.5 Pro | OpenAI | OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $30.00verified | $180.00verified |
| GPT-5.4 | OpenAI | Codex CLI, Cursor, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $2.50verified | $15.00verified |
| GPT-5.4 Mini | OpenAI | Codex CLI, Cursor, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.75verified | $4.50verified |
| GPT-5.4 Nano | OpenAI | Cursor, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.20verified | $1.25verified |
| GPT-5.3 Codex Spark | OpenAI | Codex CLI, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $1.75verified | $14.00verified |
| Gemini 3.6 Flash | Cursor, OpenCode Zen | Sonnet | Not reported | Not reported | 78.0%secondary | Not reported | 84.9verified | 20%low coverage | $1.50verified | $7.50verified | |
| Gemini 3.5 Flash | Cursor, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $1.50verified | $9.00verified | |
| Gemini 3.5 Flash Lite | OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.30verified | $2.50verified | |
| Gemini 3.1 Pro | Cursor, OpenCode Zen | Sonnet | 80.6%self-reported | Not reported | Not reported | 94.1% to 94.3%conflicted | 92.7verified | 55% | Not reported | Not reported | |
| Gemini 3 Flash | Cursor, OpenCode Zen | Light | 76.2% to 78.0%conflicted | Not reported | Not reported | Not reported | 79.4verified | 25%low coverage | Not reported | Not reported | |
| Grok 4.5 | xAI | Cursor, OpenCode Go, OpenCode Zen | Sonnet | Not reported | 64.7%self-reported | 83.3%self-reported | Not reported | 85.0verified | 45%low coverage | $2.00verified | $6.00verified |
| grok-build-0.1 | xAI | OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $1.00verified | $2.00verified |
| Cursor Composer 2.5 | Cursor | Cursor | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.50verified | $2.50verified |
| Cursor Composer 1 | Cursor | Cursor | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $1.25verified | $10.00verified |
| GLM 5.2 | Z.ai | Cursor, OpenCode Go, OpenCode Zen | Sonnet | Not reported | 62.1%secondary | 81.0% to 82.7%conflicted | 91.2%secondary | 88.0verified | 75% | $1.40verified | $4.40verified |
| GLM 5.1 | Z.ai | OpenCode Go, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $1.40verified | $4.40verified |
| Kimi K3 | Moonshot | Cursor, OpenCode Go, OpenCode Zen | Sonnet | 76.8%secondary | Not reported | 88.3%secondary | 93.5%secondary | 92.0verified | 75% | $3.00verified | $15.00verified |
| Kimi K2.7 Code | Moonshot | Cursor, OpenCode Go, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.95secondary | $4.00secondary |
| Kimi K2.6 | Moonshot | OpenCode Go, OpenCode Zen | Light | 80.2%secondary | 58.6%secondary | Not reported | Not reported | 78.3verified | 50% | $0.95verified | $4.00verified |
| MiniMax M3 | MiniMax | OpenCode Go, OpenCode Zen | Light | 80.5%secondary | 59.0%secondary | 66.0%secondary | Not reported | 76.7verified | 70% | $0.30verified | $1.20verified |
| MiniMax M2.7 | MiniMax | OpenCode Go, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.30verified | $1.20verified |
| Qwen3.7 Max | Alibaba | OpenCode Go, OpenCode Zen | Light | 80.4%secondary | Not reported | Not reported | Not reported | 83.8verified | 25%low coverage | $2.50verified | $7.50verified |
| Qwen3.7 Plus | Alibaba | OpenCode Go, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.40verified | $1.60verified |
| Qwen3.6 Plus | Alibaba | OpenCode Go, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | $0.50verified | $3.00verified |
| DeepSeek V4 Pro | DeepSeek | OpenCode Go, OpenCode Zen | Sonnet | 80.6%secondary | Not reported | Not reported | 90.1%secondary | 90.4verified | 55% | $1.74verified | $3.48verified |
| DeepSeek V4 Flash | DeepSeek | OpenCode Go, OpenCode Zen | Light | 79.0%secondary | Not reported | Not reported | Not reported | 82.3verified | 25%low coverage | $0.14verified | $0.28verified |
| MiMo v2.5 | Xiaomi | OpenCode Go, OpenCode Zen | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | Subscription ($10.00)verified | Subscription ($10.00)verified |
| MiMo v2.5 Pro | Xiaomi | OpenCode Go | Unrated | Not reported | Not reported | Not reported | Not reported | Not reported | 0% | Subscription ($10.00)verified | Subscription ($10.00)verified |
Coding-agent model roster: capability figures, composite index, coverage, and per-harness prices.