Meta, open weights
Llama 3-70B
Llama 3-70B by Meta ranks 323rd of 354 ranked models on the Noometry Index as of October 2026, with a score of 28.8. Its strongest category is agentic & tool use, where it ranks 139th.
Last verified
Specifications
- Noometry rank
- #323 of 354
- Index score
- 28.8
- Evidence
- Confirmed 31 results
- Provider
Meta- Released
- April 18, 2024
- Weights
- Open weights
- Reasoning
- Unknown
- Context window
- —
- Max output
- —
- Input price
- Not listed
- Output price
- Not listed
- Blended price
- Not listed
- Output speed
- 104 tokens/s Kagi
- Value
- Not ranked
- Knowledge cutoff
- Unknown
Category scores
Each category score combines every public result we have in that category.
- Coding 35.8
- Agentic & Tool Use 21.1
- Reasoning 18.0
- Math 12.8
- Knowledge 20.8
- Multilingual 33.6
- Instruction Following 62.5
- Long Context 35.6
- Writing & Preference 42.8
| Category | Score | Rank | Results |
|---|---|---|---|
| Coding | 35.8 | #218 | 3 |
| Agentic & Tool Use | 21.1 | #139 | 1 |
| Reasoning | 18.0 | #288 | 3 |
| Math | 12.8 | #305 | 3 |
| Knowledge | 20.8 | #277 | 2 |
| Multilingual | 33.6 | #251 | 1 |
| Instruction Following | 62.5 | #238 | 1 |
| Long Context | 35.6 | #240 | 1 |
| Writing & Preference | 42.8 | #231 | 3 |
Strengths and weaknesses
Categories where Llama 3-70B places highest and lowest among the models ranked in each, with its score against that category's median.
Strongest categories
| Category | Score | vs median | Rank |
|---|---|---|---|
| Coding | 35.8 | −2.9 | #218 of 340, top 65% |
| Writing & Preference | 42.8 | −10.9 | #231 of 312, top 75% |
| Instruction Following | 62.5 | −8.7 | #238 of 305, top 79% |
Weakest categories
| Category | Score | vs median | Rank |
|---|---|---|---|
| Math | 12.8 | −23.7 | #305 of 327, top 94% |
| Agentic & Tool Use | 21.1 | −9.2 | #139 of 154, top 91% |
| Knowledge | 20.8 | −16.5 | #277 of 314, top 89% |
Closest competitors
The models ranked just above and below Llama 3-70B. When scores are this close, price and speed are often the better way to choose.
| Model | Rank | Score | Blended $/M | Speed | |
|---|---|---|---|---|---|
| Claude 3 Sonnet | #319 | 29.0 | — | — | Compare |
| Qwen2.5 7B Instruct | #320 | 29.0 | $0.31 | — | Compare |
| Llama 3.2 3B | #321 | 28.9 | $0.12 | — | Compare |
| Qwen1.5 4b Chat | #322 | 28.8 | — | — | Compare |
| GPT-4o | #324 | 28.6 | $4.38 | — | Compare |
| Ministral 8B | #325 | 28.2 | $0.15 | — | Compare |
| Gemma 3 4B | #326 | 28.1 | $0.05 | 72 | Compare |
| GPT-4.1 nano | #327 | 27.9 | $0.18 | 135 | Compare |
Sponsored placements are available on pages like this one. Advertise on Noometry
Benchmark results
Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.
Coding
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| BigCodeBench Instruct | 43.6% | #25 of 64, top 40% | BigCodeBench | 2024-04-18 | |
| LMArena Coding | 1206 | #239 of 294, top 82% | LMArena | 2026-10-08 | |
| BigCodeBench Complete | 43.3% | BigCodeBench | 2024-04-18 | ||
| BigCodeBench Complete | 54.5% | #21 of 66, top 32% | BigCodeBench | 2024-04-18 | |
| HumanEval+ | 72% | #17 of 45, top 38% | EvalPlus | ||
| MBPP+ | 69% | #16 of 38, top 43% | EvalPlus |
Agentic & Tool Use
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| Cybench | 5% | #21 of 21, top 100% | Epoch AI |
Reasoning
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| Kagi LLM Benchmark | 35.1% | #88 of 99, top 89% | Kagi LLM Benchmark | ||
| LMArena Hard Prompts | 1195 | #239 of 297, top 81% | LMArena | 2026-10-08 | |
| DTBench | 54.2% | #125 of 151, top 83% | Epoch AI | ||
| Epoch Capabilities Index | 122.93 | #161 of 213, top 76% | Epoch AI | 2024-04-18 | |
| ForecastBench | 57.1 | #61 of 72, top 85% | Epoch AI | ||
| WinoGrande | 83.5% | #9 of 43, top 21% | Epoch AI |
Math
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 4.3% | #152 of 173, top 88% | Epoch AI | 2025-02-25 | |
| LMArena Math | 1218 | #225 of 285, top 79% | LMArena | 2026-10-08 | |
| MATH Level 5 | 22.6% | #62 of 79, top 79% | Epoch AI | 2025-01-27 |
Knowledge
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| GPQA Diamond | 40.6% | #150 of 186, top 81% | Epoch AI | 2025-01-27 | |
| LMArena Expert | 1149 | #235 of 273, top 87% | LMArena | 2026-10-08 | |
| MMLU | 79.3% | #20 of 81, top 25% | Epoch AI |
Multilingual
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Non-English | 1142 | #251 of 297, top 85% | LMArena | 2026-10-08 | |
| LMArena Chinese | 1114 | #255 of 285, top 90% | LMArena | 2026-10-08 | |
| LMArena French | 1232 | #188 of 223, top 85% | LMArena | 2026-10-08 | |
| LMArena German | 1169 | #199 of 231, top 87% | LMArena | 2026-10-08 | |
| LMArena Japanese | 1017 | #198 of 211, top 94% | LMArena | 2026-10-08 | |
| LMArena Korean | 1017 | #198 of 213, top 93% | LMArena | 2026-10-08 | |
| LMArena Russian | 1159 | #247 of 283, top 88% | LMArena | 2026-10-08 | |
| LMArena Spanish | 1241 | #187 of 226, top 83% | LMArena | 2026-10-08 |
Instruction Following
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Instruction Following | 1194 | #240 of 298, top 81% | LMArena | 2026-10-08 |
Long Context
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Longer Query | 1174 | #249 of 291, top 86% | LMArena | 2026-10-08 |
Writing & Preference
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| LMArena Text | 1221 | #237 of 297, top 80% | LMArena | 2026-10-08 | |
| LMArena Creative Writing | 1210 | #228 of 295, top 78% | LMArena | 2026-10-08 | |
| LMArena Multi-Turn | 1223 | #230 of 295, top 78% | LMArena | 2026-10-08 |
Compare Llama 3-70B
- Llama 3-70B vs Llama 2-13B
- Llama 3-70B vs Qwen1.5 4b Chat
- Llama 3-70B vs GPT-4o
- Llama 3-70B vs Llama 3.2 3B
- Llama 3-70B vs Ministral 8B
- Llama 3-70B vs Qwen2.5 7B Instruct
- Llama 3-70B vs Gemma 3 4B
- Llama 3-70B vs GPT-6 Astra
- Llama 3-70B vs Claude Fable 5.1
- Llama 3-70B vs Gemini 3.8 Flash
- Llama 3-70B vs Kimi K3
- Llama 3-70B vs Grok 4.6
- Llama 3-70B vs Qwen3.8 Max
- Llama 3-70B vs GLM-5.3
Other Meta models
- Muse Spark 1.354.8
- Muse Spark50.6
- Muse Spark 1.250.3
- Muse Spark 1.149.9
- Muse Glimmer41.7
- Codellama 70b Instruct33.7
- Llama 4 Maverick30.9
- Codellama 34b Instruct30.8
Frequently asked questions
How good is Llama 3-70B?
Llama 3-70B by Meta ranks 323rd of 354 ranked models on the Noometry Index as of October 2026, with a score of 28.8. Its strongest category is agentic & tool use, where it ranks 139th.
Is Llama 3-70B open source?
Yes. Llama 3-70B's weights are downloadable; check the license for commercial terms.
How fast is Llama 3-70B?
Llama 3-70B generated about 104 output tokens per second in the Kagi LLM Benchmark's timed runs. Speed varies by provider, load and reasoning effort.
What are Llama 3-70B's strengths and weaknesses?
Relative to other ranked models, Llama 3-70B places best in coding, writing & preference, instruction following and lowest in math, agentic & tool use, knowledge.
What is Llama 3-70B best at?
Its best category is agentic & tool use, where it ranks 139th on Noometry.