Models
Every model Solvency tracks
25 models, 16 current. Each price is verified against the provider's own pricing page on the date shown. 17 have a published pass rate; the rest are shown as missing, never estimated.
| Model | Provider | Status | In $/M | Out $/M | Cached $/M | Pass | Basis | $ / solved (heavy) | Verified |
|---|---|---|---|---|---|---|---|---|---|
| Claude Fable 5 | anthropic | current | $10 | $50 | $1 | 67% | measured | $17.46 | 2026-08-21 |
| Claude Haiku 4.5 | anthropic | current | $1 | $5 | $0.1 | missing | — | missing | 2026-08-21 |
| Claude Opus 5 | anthropic | current | $5 | $25 | $0.5 | 68% | measured | $12.01 | 2026-08-21 |
| Claude Sonnet 5 | anthropic | current | $2 | $10 | $0.2 | missing | — | missing | 2026-08-21 |
| DeepSeek V4 Flash | deepseek | current | $0.44 | $1.32 | $0.014 | 50% | measured | $0.12 | 2026-08-21 |
| DeepSeek V4 Pro | deepseek | current | $1.32 | $3.96 | $0.044 | missing | — | missing | 2026-08-21 |
| Gemini 3.1 Pro (preview) | current | $2 | $12 | $0.2 | 46% | modelled | $4.30 | 2026-08-21 | |
| Gemini 3.7 Flash | current | $0.75 | $3.75 | $0.075 | 60% | measured | $2.12 | 2026-08-21 | |
| Mistral Medium 3.5 | mistral | current | $1.5 | $7.5 | missing | missing | — | missing | 2026-08-21 |
| GPT-5.3 Codex | openai | current | $1.75 | $14 | $0.175 | missing | — | missing | 2026-08-21 |
| GPT-5.4 | openai | current | $2.5 | $15 | $0.25 | 59% | modelled | $4.19 | 2026-08-21 |
| GPT-5.6 Luna | openai | current | $0.2 | $1.2 | $0.02 | missing | — | missing | 2026-08-21 |
| GPT-5.6 Sol | openai | current | $5 | $30 | $0.5 | 65% | measured | $9.88 | 2026-08-21 |
| GPT-5.6 Terra | openai | current | $2 | $12 | $0.2 | missing | — | missing | 2026-08-21 |
| Grok 4.5 | xai | current | $2 | $6 | $0.3 | 64% | measured | $3.81 | 2026-08-21 |
| Grok 4.6 | xai | current | $2 | $6 | $0.5 | missing | — | missing | 2026-08-21 |
| Claude Opus 4.5 | anthropic | legacy | $5 | $25 | $0.5 | 46% | modelled | $9.81 | 2026-08-21 |
| Claude Opus 4.6 | anthropic | legacy | $5 | $25 | $0.5 | 52% | modelled | $8.67 | 2026-08-21 |
| Gemini 2.5 Pro | legacy | $1.25 | $10 | $0.125 | 83% | modelled · stale | $1.76 | 2026-08-21 | |
| GPT-4.1 | openai | legacy | $2 | $8 | $0.5 | 52% | modelled · stale | $5.15 | 2026-08-21 |
| GPT-5 | openai | legacy | $1.25 | $10 | $0.125 | 88% | modelled · stale | $1.66 | 2026-08-21 |
| o3 | openai | legacy | $2 | $8 | $0.5 | 81% | modelled · stale | $1.99 | 2026-08-21 |
| o3-pro | openai | legacy | $20 | $80 | missing | 85% | modelled · stale | $19.08 | 2026-08-21 |
| Claude Opus 4 | anthropic | retired | $15 | $75 | $1.5 | 72% | modelled · stale | $18.75 | 2026-08-21 |
| Claude Sonnet 4 | anthropic | retired | $3 | $15 | $0.3 | 61% | modelled · stale | $4.40 | 2026-08-21 |
verified 2026-08-21 · Measured and modelled figures in the last column are on different bases and must not be ranked against each other; see the leaderboard for the separated ranking.