- Input /1M
- $0.02
- Output /1M
- $0.04
- Reference cost
- $0.03
- Context
- 131K
- AA Intelligence
- 6.9
AI model capabilities
AI models that support log probabilities
Compare models that expose token-level log probabilities for evaluation and analysis. Compare 125 supporting models by input price, output price, context window and published intelligence. Find the cheapest option for a reference text workload, then check the serving API before you build.
125 supporting models20 AI providersUSD · prices per 1M tokens
Lowest reference cost
Llama 3.1 8B Instruct$0.03
1M input + 250K output tokens. Tool fees and discounts excluded.
Highest published intelligence
Grok 4.746.4 AA Intelligence Index
Among supporting models with a published score; not a capability-specific benchmark.
Cheapest AI models with log probabilities
Ranked by the cost of 1 million input and 250,000 output tokens at standard rates. A different token mix can change the cheapest model. Support requires an explicitly listed catalog feature, not an inference from a model family.
| Model / provider | Input /1M | Output /1M | Reference cost | Context | AA Intelligence |
|---|---|---|---|---|---|
| Llama 3.1 8B InstructMeta | $0.02 | $0.04 | $0.03 | 131K | 6.9 |
| Mistral NemoMistral AI | $0.03 | $0.03 | $0.04 | 131K | — |
| Ling 3.0 FlashinclusionAI | $0.02 | $0.06 | $0.04 | 262K | — |
| Nex-N2.5-MiniNex AGI | $0.03 | $0.10 | $0.05 | 262K | — |
| GLM 5.3 FlashZ.ai | $0.03 | $0.08 | $0.05 | 1M | 41.8 |
| gpt-oss-20bOpenAI | $0.03 | $0.13 | $0.06 | 131K | 9 |
| Qwen3.7 FlashQwen | $0.03 | $0.13 | $0.06 | 1M | — |
| Ling 3.0 Flash FininclusionAI | $0.04 | $0.12 | $0.07 | 262K | — |
| Ling 3.0 Flash SanteinclusionAI | $0.04 | $0.12 | $0.07 | 262K | — |
| Gemma 3 4BGoogle | $0.05 | $0.10 | $0.08 | 131K | 4.8 |
| Ling 3.1 FlashinclusionAI | $0.04 | $0.14 | $0.08 | 1M | — |
| DeepSeek V4 Flash 0731DeepSeek | $0.0065 | $0.30 | $0.08 | 1M | — |
| Nemotron 3.5 LightningNVIDIA | $0.07 | $0.20 | $0.12 | 262K | 12.9 |
| Llama 3.2 3B InstructMeta | $0.05 | $0.33 | $0.13 | 131K | 5.7 |
| DeepSeek V4 Flash 0423DeepSeek | $0.09 | $0.18 | $0.14 | 1M | 34.3 |
| Nex-N2.5-ProNex AGI | $0.07 | $0.25 | $0.14 | 262K | — |
| Qwen3 32BQwen | $0.08 | $0.28 | $0.15 | 131K | — |
| GLM 4.7 FlashZ.ai | $0.06 | $0.40 | $0.16 | 200K | 14.9 |
| Gemma 4 26B A4BGoogle | $0.09 | $0.30 | $0.16 | 262K | 16.7 |
| Mistral Small 3.2 24BMistral AI | $0.10 | $0.30 | $0.17 | 256K | — |
| Nemotron 3 SuperNVIDIA | $0.07 | $0.40 | $0.17 | 1M | — |
| Qwen3 30B A3B Instruct 2507Qwen | $0.10 | $0.30 | $0.17 | 262K | 7.5 |
| Qwen3 14BQwen | $0.12 | $0.24 | $0.18 | 41K | — |
| Gemma 3 27BGoogle | $0.08 | $0.45 | $0.19 | 131K | 4.9 |
| DeepSeek V4.1 FlashDeepSeek | $0.05 | $0.57 | $0.19 | 1M | 39.5 |
| Qwen3 VL 32B InstructQwen | $0.10 | $0.42 | $0.21 | 131K | 8.4 |
| MiMo-V2.6-FlashXiaomi | $0.14 | $0.28 | $0.21 | 1.1M | 37.9 |
| LFM2.5-24B-A2BLiquid | $0.05 | $0.70 | $0.22 | 128K | — |
| Qwen3.5-9BQwen | $0.17 | $0.25 | $0.23 | 262K | 11.2 |
| Qwen3.8 FlashQwen | $0.15 | $0.47 | $0.27 | 1M | — |
| Qwen3.8 Omni FlashQwen | $0.15 | $0.47 | $0.27 | 1M | — |
| DeepSeek V4.1 Flash 1017DeepSeek | $0.15 | $0.50 | $0.28 | 1M | — |
| Nemotron 3.5 Super VLNVIDIA | $0.15 | $0.50 | $0.28 | 1M | — |
| GPT-4o miniOpenAI | $0.15 | $0.60 | $0.30 | 128K | 6.7 |
| GPT-4o-mini (2024-07-18)OpenAI | $0.15 | $0.60 | $0.30 | 128K | — |
| gpt-oss-120bOpenAI | $0.15 | $0.60 | $0.30 | 131K | 11.6 |
| Qwen3 VL 30B A3B InstructQwen | $0.15 | $0.60 | $0.30 | 262K | 7.9 |
| Gemma 4 31BGoogle | $0.20 | $0.40 | $0.30 | 262K | 14.7 |
| DeepSeek V4 Pro 0423DeepSeek | $0.21 | $0.42 | $0.31 | 1M | 36 |
| Qwen3 Coder NextQwen | $0.12 | $0.80 | $0.32 | 262K | 9.2 |
| Qwen3.6 35B A3BQwen | $0.10 | $0.90 | $0.33 | 262K | 18.2 |
| Qwen3 235B A22B Instruct 2507Qwen | $0.14 | $0.80 | $0.34 | 262K | — |
| DeepSeek V3.2DeepSeek | $0.26 | $0.38 | $0.35 | 164K | 16 |
| Qwen3 Next 80B A3B InstructQwen | $0.09 | $1.10 | $0.36 | 262K | 9.6 |
| Qwen3.5-35B-A3BQwen | $0.14 | $1.00 | $0.39 | 262K | 19.3 |
| Qwen3 VL 8B InstructQwen | $0.25 | $0.75 | $0.44 | 262K | 7.3 |
| Qwen3 Coder FlashQwen | $0.20 | $0.97 | $0.44 | 1M | — |
| Qwen3 Next 80B A3B ThinkingQwen | $0.15 | $1.20 | $0.45 | 262K | — |
| Qwen Plus 0728Qwen | $0.26 | $0.78 | $0.46 | 1M | — |
| Qwen-PlusQwen | $0.26 | $0.78 | $0.46 | 1M | — |
| Qwen3.6 FlashQwen | $0.19 | $1.13 | $0.47 | 1M | — |
| DeepSeek V3.1DeepSeek | $0.25 | $0.95 | $0.49 | 164K | 13.7 |
| Step 3.7 FlashStepFun | $0.20 | $1.15 | $0.49 | 262K | — |
| Mistral Small 3.1 24BMistral AI | $0.35 | $0.56 | $0.49 | 128K | — |
| DeepSeek V3 0324DeepSeek | $0.29 | $1.14 | $0.57 | 164K | 9.7 |
| Weaver (alpha)Mancer 2 | $0.40 | $0.75 | $0.59 | 8K | — |
| Llama 4 MaverickMeta | $0.35 | $1.00 | $0.60 | 524K | 10 |
| MiniMax M2.5MiniMax | $0.30 | $1.20 | $0.60 | 200K | 22.8 |
| MiniMax M3MiniMax | $0.30 | $1.20 | $0.60 | 1M | 29.2 |
| Qwen3.8 27BQwen | $0.15 | $1.88 | $0.62 | 1M | 33.7 |
| Qwen3.7 PlusQwen | $0.32 | $1.28 | $0.64 | 1M | 25.2 |
| Qwen3.5 Plus 2026-02-15Qwen | $0.26 | $1.56 | $0.65 | 1M | — |
| MiMo-V2.6-ProXiaomi | $0.43 | $0.87 | $0.65 | 1.1M | 46.3 |
| Qwen3 VL 235B A22B InstructQwen | $0.21 | $1.90 | $0.68 | 262K | 9.9 |
| Qwen3 VL 8B ThinkingQwen | $0.18 | $2.10 | $0.71 | 131K | — |
| Qwen3.5 Plus 2026-04-20Qwen | $0.30 | $1.80 | $0.75 | 1M | — |
| DeepSeek V4 Flash Vision ExpDeepSeek | $0.44 | $1.32 | $0.77 | 1M | 34.8 |
| Qwen3.5-122B-A10BQwen | $0.26 | $2.08 | $0.78 | 262K | 15.6 |
| Llama 3.3 70B InstructMeta | $0.59 | $0.79 | $0.79 | 131K | 7.7 |
| Qwen3.6 PlusQwen | $0.33 | $1.95 | $0.81 | 1M | 27 |
| GPT-3.5 TurboOpenAI | $0.50 | $1.50 | $0.88 | 16K | — |
| Qwen3.5-27BQwen | $0.26 | $2.60 | $0.91 | 262K | 22.9 |
| MiMo-V2.5-ProXiaomi | $0.48 | $1.80 | $0.93 | 1M | 26 |
| R1 0528DeepSeek | $0.50 | $2.15 | $1.04 | 164K | — |
| GLM 5.3Z.ai | $0.70 | $1.40 | $1.05 | 1M | 44.8 |
| Qwen2.5 VL 72B InstructQwen | $0.80 | $1.00 | $1.05 | 128K | — |
| Morph V3 FastMorph | $0.80 | $1.20 | $1.10 | 262K | — |
| Qwen3.6 27BQwen | $0.32 | $3.20 | $1.12 | 262K | 21.4 |
| GPT Audio MiniOpenAI | $0.60 | $2.40 | $1.20 | 128K | — |
| Kimi K2 0905MoonshotAI | $0.60 | $2.50 | $1.23 | 262K | 15.3 |
| Mistral Large 4Mistral AI | $0.70 | $2.10 | $1.23 | 1M | — |
| Hy4 previewTencent | $0.75 | $2.25 | $1.31 | 1M | — |
| Kimi K2.5MoonshotAI | $0.53 | $3.33 | $1.36 | 262K | 23.5 |
| Morph V3 LargeMorph | $0.90 | $1.90 | $1.38 | 262K | — |
| Qwen3 VL 235B A22B ThinkingQwen | $0.40 | $4.00 | $1.40 | 131K | — |
| Qwen3 Coder PlusQwen | $0.65 | $3.25 | $1.46 | 1M | — |
| Kimi K2.7 CodeMoonshotAI | $0.60 | $3.50 | $1.48 | 262K | 25.8 |
| GPT-3.5 Turbo (older v0613)OpenAI | $1.00 | $2.00 | $1.50 | 4K | — |
| Grok Build 0.1xAI | $1.00 | $2.00 | $1.50 | 256K | — |
| Qwen3.5 397B A17BQwen | $0.60 | $3.60 | $1.50 | 262K | 18.4 |
| Kimi K2.6MoonshotAI | $0.65 | $3.41 | $1.50 | 262K | 27 |
| Step 5 PreviewStepFun | $1.00 | $2.70 | $1.68 | 1M | — |
| Qwen3 MaxQwen | $0.78 | $3.90 | $1.75 | 262K | 15.6 |
| Qwen3 Max ThinkingQwen | $0.78 | $3.90 | $1.75 | 262K | 21.3 |
| DeepSeek V4 Pro 0813DeepSeek | $0.66 | $4.44 | $1.77 | 1M | — |
| GLM 5Z.ai | $1.00 | $3.20 | $1.80 | 205K | 27.9 |
| Grok 4.20xAI | $1.25 | $2.50 | $1.88 | 2M | 25.7 |
| Grok 4.20 Multi-AgentxAI | $1.25 | $2.50 | $1.88 | 2M | — |
| Grok 4.3xAI | $1.25 | $2.50 | $1.88 | 1M | 24.9 |
| GLM 5.2Z.ai | $0.20 | $6.80 | $1.90 | 1M | 33.7 |
| GPT-3.5 Turbo InstructOpenAI | $1.50 | $2.00 | $2.00 | 4K | — |
| Muse Spark OpenMeta | $1.25 | $4.25 | $2.31 | 1M | — |
| GLM 5.1Z.ai | $1.40 | $4.40 | $2.50 | 205K | 26.1 |
| Qwen3.6 Max PreviewQwen | $1.03 | $6.16 | $2.57 | 262K | 28.4 |
| Qwen3.7 MaxQwen | $1.48 | $4.42 | $2.58 | 1M | 29.5 |
| GPT-5 Image MiniOpenAI | $2.50 | $2.00 | $3.00 | 400K | — |
| Grok 4.5xAI | $2.00 | $6.00 | $3.50 | 500K | 38.8 |
| Grok 4.6xAI | $2.00 | $6.00 | $3.50 | 500K | 44.3 |
| Grok 4.7xAI | $2.00 | $6.00 | $3.50 | 500K | 46.4 |
| Qwen3.8 2.4T A95BQwen | $2.00 | $6.00 | $3.50 | 1M | 39.9 |
| Qwen3.8 Max (0902)Qwen | $2.00 | $6.00 | $3.50 | 1M | — |
| GPT-3.5 Turbo 16kOpenAI | $3.00 | $4.00 | $4.00 | 16K | — |
| GLM 5.3 PrimeZ.ai | $2.80 | $8.80 | $5.00 | 1M | — |
| GPT AudioOpenAI | $2.50 | $10 | $5.00 | 128K | — |
| GPT-4oOpenAI | $2.50 | $10 | $5.00 | 128K | 8.4 |
| GPT-4o (2024-08-06)OpenAI | $2.50 | $10 | $5.00 | 128K | 7.7 |
| GPT-4o (2024-11-20)OpenAI | $2.50 | $10 | $5.00 | 128K | — |
| Ember-1Fireworks | $3.00 | $15 | $6.75 | 1M | — |
| Kimi K3MoonshotAI | $3.00 | $16 | $7.00 | 1M | 43.6 |
| Qwen3.8 Max PrimeQwen | $4.00 | $12 | $7.00 | 1M | — |
| GPT-4o (2024-05-13)OpenAI | $5.00 | $15 | $8.75 | 128K | 7.3 |
| GPT-5.4 Image 2OpenAI | $8.00 | $15 | $12 | 272K | — |
| GPT-5 ImageOpenAI | $10 | $10 | $13 | 400K | — |
| GPT-4 TurboOpenAI | $10 | $30 | $18 | 128K | 7 |
| GPT-4OpenAI | $30 | $60 | $45 | 8K | 6.7 |
- Input /1M
- $0.03
- Output /1M
- $0.03
- Reference cost
- $0.04
- Context
- 131K
- AA Intelligence
- —
- Input /1M
- $0.02
- Output /1M
- $0.06
- Reference cost
- $0.04
- Context
- 262K
- AA Intelligence
- —
- Input /1M
- $0.03
- Output /1M
- $0.10
- Reference cost
- $0.05
- Context
- 262K
- AA Intelligence
- —
- Input /1M
- $0.03
- Output /1M
- $0.08
- Reference cost
- $0.05
- Context
- 1M
- AA Intelligence
- 41.8
- Input /1M
- $0.03
- Output /1M
- $0.13
- Reference cost
- $0.06
- Context
- 131K
- AA Intelligence
- 9
- Input /1M
- $0.03
- Output /1M
- $0.13
- Reference cost
- $0.06
- Context
- 1M
- AA Intelligence
- —
- Input /1M
- $0.04
- Output /1M
- $0.12
- Reference cost
- $0.07
- Context
- 262K
- AA Intelligence
- —
- Input /1M
- $0.04
- Output /1M
- $0.12
- Reference cost
- $0.07
- Context
- 262K
- AA Intelligence
- —
- Input /1M
- $0.05
- Output /1M
- $0.10
- Reference cost
- $0.08
- Context
- 131K
- AA Intelligence
- 4.8
- Input /1M
- $0.04
- Output /1M
- $0.14
- Reference cost
- $0.08
- Context
- 1M
- AA Intelligence
- —
- Input /1M
- $0.0065
- Output /1M
- $0.30
- Reference cost
- $0.08
- Context
- 1M
- AA Intelligence
- —
- Input /1M
- $0.07
- Output /1M
- $0.20
- Reference cost
- $0.12
- Context
- 262K
- AA Intelligence
- 12.9
- Input /1M
- $0.05
- Output /1M
- $0.33
- Reference cost
- $0.13
- Context
- 131K
- AA Intelligence
- 5.7
- Input /1M
- $0.09
- Output /1M
- $0.18
- Reference cost
- $0.14
- Context
- 1M
- AA Intelligence
- 34.3
Showing 15 of 125
125 matching models · Missing intelligence scores are shown as —.
What is log probabilities?
Log probabilities describe how likely generated tokens were under a model’s distribution. They can help with controlled classification experiments and response analysis, but they are not calibrated probabilities that an answer is correct. Availability and the number of alternative tokens returned vary by API.
Before choosing a model
- Check whether top-token alternatives and your desired output format are supported.
- Evaluate calibration before using scores to make business decisions.
- Do not treat token likelihood as factual confidence.
Estimate your actual workload
Use your request volume and prompt and response sizes to compare monthly text-token costs. Add any search, external tool or provider-specific charges separately. Standard token prices do not assume cache hits or batch discounts.
Log probabilities FAQs
125 active models with positive input and output prices currently list log probabilities in the PriceIndex catalog. The table lists every match. A model is included only when its tracked feature data explicitly lists this capability; missing data is not treated as support.
Llama 3.1 8B Instruct has the lowest estimated text-token cost in this list for 1 million input tokens and 250,000 output tokens: $0.0300. A different input/output mix can change the ranking. This excludes tool fees, caching, discounts and taxes.
No. Capability availability can differ between serving APIs and model versions. Follow a model’s API documentation link and verify the exact version, endpoint and settings before choosing it.
We use standard USD input and output token prices: input price × 1 million tokens plus output price × 250,000 tokens, divided by 1 million. The directory excludes retired models, missing prices and zero-priced entries. Published AA Intelligence Index scores are shown when available and never estimated.
Check whether top-token alternatives and your desired output format are supported. Evaluate calibration before using scores to make business decisions. Do not treat token likelihood as factual confidence.