Across 7 comparable benchmarks, Llama3.3-70B-Instruct has better reported results on 7 and worse on 0.
Moderate evidence7 of 7 shared results countedHow we compare
Price basis: Standard rates. The price section below shows the applicable periods.
| Fact | GPT-4o mini | Llama3.3-70B-Instruct |
|---|---|---|
| Context window | 128K tokens | 128K tokens |
| Input / 1M | $0.150 | $0.100 |
| Output / 1M | $0.600 | $0.320 |
| Blended (4:1) | $0.186 | $0.144 |
| Status | Deprecated | Available |
| Released | July 18, 2024 | December 6, 2024 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 4 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
Llama3.3-70B-Instruct ahead on 1 of 1 counted
Llama3.3-70B-Instruct ahead on 2 of 2 counted
Llama3.3-70B-Instruct ahead on 1 of 1 counted
Llama3.3-70B-Instruct ahead on 1 of 1 counted
No shared comparable results for Computer & tool use, Research & search, Long context, Multimodal, Professional work & productivity, Cybersecurity.
Published scores, with matching modes where available. Gaps show GPT-4o mini relative to Llama3.3-70B-Instruct. Reading the results
| Benchmark | GPT-4o mini | Llama3.3-70B-Instruct | Gap |
|---|---|---|---|
| HumanEvalCode Generation & Editing · Accuracy | 87.2Normal · no tools | 88.4Normal · no tools | −1.2 pts#33 vs #23 of 100 |
| BALROGLogical Planning · Average progress (%) | 17.4Normal · tools | 23Normal · tools | −5.6 pts#24 vs #20 of 31 |
| SimpleBenchCommonsense · Score (AVG@5) | 10.7Normal · no tools | 19.9Normal · no tools | 0.54×#106 vs #101 of 106 |
| MATHMathematics · Accuracy | 70.2Normal · no tools | 77.0Normal · no tools | −6.8 pts#19 vs #13 of 42 |
| GPQA DiamondScientific Reasoning · Accuracy | 41.1Normal · no tools | 50.5Normal · no tools | −9.4 pts#188 vs #176 of 200 |
| MMLUKnowledge Exams · Accuracy | 82.0Normal · no tools | 86.0Normal · no tools | −4.0 pts#41 vs #29 of 114 |
| MMLU-ProKnowledge Exams · Accuracy | 63.1Reported best | 68.9≠Normal · no tools | −5.8 pts#107 vs #97 of 153 |
Code Generation & Editing · Accuracy
Logical Planning · Average progress (%)
Commonsense · Score (AVG@5)
Mathematics · Accuracy
Scientific Reasoning · Accuracy
A dash means no result is available.
| Benchmark | GPT-4o mini | Llama3.3-70B-Instruct |
|---|---|---|
| GPQA | 40.2 | — |
| GSM8K | 91.3 | — |
| Aider-Polyglot | 3.6 | — |
| LiveCodeBench | — | 33.3 |
| MBPP | — | 87.6 |
| WeirdML v2 | 11.8 | — |
| Pinch Bench | 75.0 | — |
| ECI | 127 | — |
| Creative Writing | 871 | — |
| Fiction.liveBench | — | 33.3 |
| GeoBench ACW | 64.0 | — |
| MMMU | 59.4 | — |
| MMMU-Pro | 41.5 | — |
| VPCT | 34.0 | — |
| τ³-Banking | 2.9 | — |
| SimpleQA | 9.5 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
GPT-4o mini costs29%morethan Llama3.3-70B-Instruct per million tokens.Using standard rates at this token mix.
GPT-4o mini: Standard pricing. OpenAI official pricing (developers.openai.com/api/docs/pricing) verified 2026-11-13; standard tier; Deprecated; shutdown 2027-01-14 (replacement: gpt-6-luna)
Llama3.3-70B-Instruct: Standard pricing. DeepInfra official models API (meta-llama/Llama-3.3-70B-Instruct-Turbo) verified 2026-11-13
| Price / 1M tokens | GPT-4o mini | Llama3.3-70B-Instruct | Ratio |
|---|---|---|---|
| Input | $0.15 | $0.10 | 1.50× |
| Output | $0.60 | $0.32 | 1.88× |
| Cached input | $0.075 | — | — |
| Cache write | — | — | — |
| Blended | $0.186 | $0.144 | 1.29× |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
| Attribute | GPT-4o mini | Llama3.3-70B-Instruct |
|---|---|---|
| Availability | Deprecated | Available |
| Provider | OpenAI | Facebook AI Research Lab |
| Model type | Foundation model | Chat model |
| First release | July 18, 2024 | December 6, 2024 |
| Context window | 128K tokens | 128K tokens |
| Max output | 16K tokens | — |
| Total parameters | — | 70B |
| Architecture | Undisclosed | Dense |
| Documented capabilities | Multilingual | — |
| Input modalities | Text, Image | Text |
| Output modalities | Text | Text |
| Weights | Closed / API only | Open weights |
| Licence | — | LLAMA 3.1 COMMUNITY LICENSE AGREEMENT |
| Commercial use | Proprietary | Commercial use allowed |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used GPT-4o mini or Llama3.3-70B-Instruct? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.