MetaLlama 3.1 70B
Best configuration ranks #502 of 610 on the BenchLeader Index at 40.6 ±4.5. Last measured 2 Sept 2026. Released 16 Jul 2024.
- Blended price
- $0.560/M
- $0.560 in · $0.560 out
- Output speed
- 63 tok/s
- First answer
- 1.60 s
- first token 0.36 s
- Context
- 128k
- Overall index41
- Reasoning38
- Coding34
- Agents & tools36
- Maths33
- Knowledge44
- Instruction following38
- Human preference45
- Composite37
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 | 44.2%#221 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1298#240 | LMArena |
| GPQA Diamond (AA)not in index | 40.9%#450 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.5%#422 | Artificial Analysis |
Coding
| Benchmark | default | Source |
|---|---|---|
| WeirdML | 9.0%#146 | WeirdML |
| LMArena Coding | 1333#235 | LMArena |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 3.0%#298 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 15.2%#346 | Artificial Analysis |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 3.6%#233 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 36.7%#64 | Epoch AI Benchmarking Hub |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -43.1#274 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 34.4%#303 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1293#236 | LMArena |
Composite
| Benchmark | default | Source |
|---|---|---|
| Epoch Capabilities Indexnot in index | 125.9#148 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 6.6#436 | 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 |
|---|---|---|---|---|---|---|
| Amazon Bedrock | 20 tok/s | 0.38 s | $0.720 | $0.720 | 131k | – |
| DeepInfra (Turbo) | 16 tok/s | 0.35 s | $0.400 | $0.400 | 131k | fp8 |
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 | $0.0004 | – | 6.4 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0071 | – | 11.2 s |
| Code edit | 6,000 / 1,500 | $0.0042 | – | 25.6 s |
| Agentic coding session | 60,000 / 4,000 | $0.036 | – | 1.1 min |
| Structured extraction | 2,000 / 200 | $0.0012 | – | 4.8 s |
See also
Data as of 9 Sept 2026. Compare with another model.