OpenAI
gpt-oss-120b
$0.065 / 1M tokens
cheapest at W&B Inference (CoreWeave) · $0.03 in · $0.17 out
Wins 2 of 4
Head to head · prices updated Oct 31, 2026
gpt-oss-120b vs Llama 4 Maverick 17B 128E Instruct FP8
Llama 4 Maverick 17B 128E Instruct FP8 is 4% cheaper than gpt-oss-120b at their cheapest providers (blended 3 input : 1 output tokens).
Meta Llama
Llama 4 Maverick 17B 128E Instruct FP8
$0.0625 / 1M tokens
cheapest at Lambda · $0.05 in · $0.10 out
Wins 2 of 4
- Lower priceLlama 4 Maverick 17B 128E Instruct FP8
- Higher accuracy (GPQA)gpt-oss-120b
- Longer contextLlama 4 Maverick 17B 128E Instruct FP8
- More providersgpt-oss-120b
Verdict
Llama 4 Maverick 17B 128E Instruct FP8 is the better deal on price
Llama 4 Maverick 17B 128E Instruct FP8 is 4% cheaper than gpt-oss-120b at their cheapest providers (blended 3 input : 1 output tokens).
gpt-oss-120b scores higher on GPQA Diamond: 75.8% (high reasoning) vs 67.0%.
So the choice is a trade-off: gpt-oss-120b for the hardest reasoning, Llama 4 Maverick 17B 128E Instruct FP8 when cost matters more than the last few points of accuracy.
Llama 4 Maverick 17B 128E Instruct FP8 accepts longer prompts (1049K vs 131K tokens).
Your monthly bill
What would each cost you?
gpt-oss-120b saves you $0.06 a month ($0.72 a year) for this workload.
Side by side
| gpt-oss-120b | Llama 4 Maverick 17B 128E Instruct FP8 | |
|---|---|---|
| Cheapest blended price / 1M | $0.065 | $0.0625 |
| Input / output per 1M | $0.03 / $0.17 | $0.05 / $0.10 |
| Accuracy (GPQA Diamond) | 75.8% (high reasoning) | 67.0% |
| Max context | 131K | 1M |
| Providers we track | 18 | 6 |
Standard list prices at each model's cheapest provider we track; blended = 3 input : 1 output tokens. GPQA Diamond from Epoch AI (CC BY 4.0), best measured reasoning setting, named next to each score. Methodology.
Where to buy each one
gpt-oss-120b
- W&B Inference (CoreWeave)$0.065
- DeepInfra$0.0703
- Novita AI$0.10
- SiliconFlow$0.15
- OVHcloud AI Endpoints$0.1788
Llama 4 Maverick 17B 128E Instruct FP8
- Lambda$0.0625
- DeepInfra$0.35
- Novita AI$0.415
- Microsoft Azure AI Foundry$0.4375
- IBM watsonx.ai$0.6493
Questions people ask
Is gpt-oss-120b cheaper than Llama 4 Maverick 17B 128E Instruct FP8?
Llama 4 Maverick 17B 128E Instruct FP8 is 4% cheaper than gpt-oss-120b at their cheapest providers (blended 3 input : 1 output tokens).
Which is more accurate, gpt-oss-120b or Llama 4 Maverick 17B 128E Instruct FP8?
On GPQA Diamond (graduate-level science questions, measured by Epoch AI) gpt-oss-120b scores 75.8% (high reasoning) and Llama 4 Maverick 17B 128E Instruct FP8 67.0%. One benchmark is not the whole picture; test both on your own prompts.
How much does gpt-oss-120b cost per 1M tokens?
From $0.03 input and $0.17 output per 1M tokens at W&B Inference (CoreWeave), the cheapest of 18 providers we track.
How much does Llama 4 Maverick 17B 128E Instruct FP8 cost per 1M tokens?
From $0.05 input and $0.10 output per 1M tokens at Lambda, the cheapest of 6 providers we track.
Swap a model