MetaLlama 3-70B
Best configuration ranks #544 of 610 on the BenchLeader Index at 38.6 ±3.1. Last measured 2 Sept 2026. Released 18 Apr 2024.
- Blended price
- $1.18/M
- $0.650 in · $2.75 out
- Output speed
- –
- First answer
- –
- Context
- 8k
- Overall index39
- Reasoning35
- Coding42
- Agents & tools35
- Maths30
- Knowledge38
- Instruction following40
- Human preference43
- Long context25
- Composite36
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 | 40.6%#230 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1279#251 | LMArena |
| GPQA Diamond (AA)not in index | 37.9%#460 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.5%#418 | Artificial Analysis |
| Kagi LLM Benchmark | 35.1%#117 | Kagi LLM Benchmark |
Coding
| Benchmark | default | Source |
|---|---|---|
| LMArena Coding | 1306#251 | LMArena |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Cybench | 5.0%#21 | Cybench |
| Terminal-Bench Hard | 0.8%#332 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 0.0%#375 | Artificial Analysis |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 4.3%#229 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 22.6%#72 | Epoch AI Benchmarking Hub |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -54.3#353 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 37.1%#279 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1276#250 | LMArena |
Long context
| Benchmark | default | Source |
|---|---|---|
| AA-LCR | 0.0%#419 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| Epoch Capabilities Indexnot in index | 122.9#155 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 5.5#491 | Artificial Analysis |
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 | $0.0011 | – | – |
| Summarise a 30-page report | 12,000 / 600 | $0.0095 | – | – |
| Code edit | 6,000 / 1,500 | $0.0080 | – | – |
| Agentic coding session | 60,000 / 4,000 | $0.050 | – | – |
| Structured extraction | 2,000 / 200 | $0.0019 | – | – |
See also
Data as of 9 Sept 2026. Compare with another model.