Compare · ModelsLive · 2 picked · head to head
Claude 3 Haiku vs Llama 3 70B Instruct
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Llama 3 70B Instruct wins on 5/5 benchmarks
Llama 3 70B Instruct wins 5 of 5 shared benchmarks. Leads in knowledge · math.
Category leads
knowledge·Llama 3 70B Instructmath·Llama 3 70B Instruct
Hype vs Reality
Attention vs performance
Claude 3 Haiku
#191 by perf·no signal
Llama 3 70B Instruct
#181 by perf·no signal
Best value
Llama 3 70B Instruct
1.4x better value than Claude 3 Haiku
Claude 3 Haiku
38.3 pts/$
$0.75/M
Llama 3 70B Instruct
51.8 pts/$
$0.63/M
Vendor risk
Who is behind the model
Anthropic
$380.0B·Tier 1
Meta AI
$1.50T·Tier 1
Head to head
5 benchmarks · 2 models
Claude 3 HaikuLlama 3 70B Instruct
GPQA diamond
Llama 3 70B Instruct leads by +5.7
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Claude 3 Haiku
15.1
Llama 3 70B Instruct
20.8
MATH level 5
Llama 3 70B Instruct leads by +7.7
MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics.
Claude 3 Haiku
14.9
Llama 3 70B Instruct
22.6
MMLU
Llama 3 70B Instruct leads by +7.3
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
Claude 3 Haiku
65.1
Llama 3 70B Instruct
72.4
OTIS Mock AIME 2024-2025
Llama 3 70B Instruct leads by +2.5
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Claude 3 Haiku
1.7
Llama 3 70B Instruct
4.2
Winogrande
Llama 3 70B Instruct leads by +18.6
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
Claude 3 Haiku
48.4
Llama 3 70B Instruct
67.0
Full benchmark table
| Benchmark | Claude 3 Haiku | Llama 3 70B Instruct |
|---|---|---|
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 15.1 | 20.8 |
MATH level 5 MATH Level 5 · the hardest tier of the MATH benchmark, featuring competition-level problems from AMC, AIME, and Olympiad-style mathematics. | 14.9 | 22.6 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 65.1 | 72.4 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 1.7 | 4.2 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 48.4 | 67.0 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.25 | $1.25 | 200K tokens (~100 books) | $5.00 | |
| $0.51 | $0.74 | 8K tokens (~4 books) | $5.67 |