MetaLlama 3.1 8B
Best configuration ranks #574 of 610 on the BenchLeader Index at 36.6 ±5.3. Last measured 2 Sept 2026. Released 1 Jan 2025.
- Blended price
- $0.053/M
- $0.050 in · $0.060 out
- Output speed
- 142 tok/s
- First answer
- 0.89 s
- first token 0.50 s
- Context
- 128k
- Overall index37
- Reasoning27
- Coding21
- Agents & tools37
- Maths29
- Knowledge49
- Instruction following33
- Human preference35
- Long context34
- Composite38
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source |
|---|---|---|
| GPQA Diamond | 27.0%#259 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1222#282 | LMArena |
| GPQA Diamond (AA)not in index | 25.9%#508 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 5.3%#359 | Artificial Analysis |
Coding
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 0.8%#332 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 16.4%#339 | Artificial Analysis |
| BFCL Overall | 25.8%#64 | Berkeley Function Calling Leaderboard |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 1.7%#244 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 22.9%#70 | Epoch AI Benchmarking Hub |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -30.9#219 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 28.6%#353 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1211#283 | LMArena |
Long context
| Benchmark | default | Source |
|---|---|---|
| AA-LCR | 18.0%#371 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| Epoch Capabilities Indexnot in index | 116.5#177 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 6.9#422 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| CoreWeave | 105 tok/s | 0.24 s | $0.220 | $0.220 | 131k | bf16 |
| Groq | 84 tok/s | 0.29 s | $0.050 | $0.080 | 131k | – |
| NovitaAI | 66 tok/s | 0.50 s | $0.020 | $0.050 | 16k | fp8 |
| DeepInfra | 17 tok/s | 0.69 s | $0.020 | $0.040 | 131k | fp8 |
| Cloudflare | 15 tok/s | 0.50 s | $0.152 | $0.287 | 32k | fp8 |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0000 | – | 3.0 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0006 | – | 5.1 s |
| Code edit | 6,000 / 1,500 | $0.0004 | – | 11.5 s |
| Agentic coding session | 60,000 / 4,000 | $0.0032 | – | 29.1 s |
| Structured extraction | 2,000 / 200 | $0.0001 | – | 2.3 s |
See also
Data as of 9 Sept 2026. Compare with another model.