Model comparison
GPT-5.5 Instant vs Llama-3.3-70B-Instruct
GPT-5.5 Instant is the stronger model overall, scoring 42.5 to 30.3 on the Noometry Index.
Last verified . 22 shared benchmarks.
MetaSummary
- They share 22 benchmarks with published results for both. GPT-5.5 Instant scores higher in 8 categories and Llama-3.3-70B-Instruct in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5.5 Instant leads 48.9 to 30.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 68.1% for GPT-5.5 Instant and 5.1% for Llama-3.3-70B-Instruct.
- Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.
Side by side
| GPT-5.5 Instant | Llama-3.3-70B-Instruct | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 42.5 | 30.3 |
| Released | 2026-05-05 | 2024-12-06 |
| Weights | Proprietary | Open |
| Context window | — | 128K |
| Max output | — | 4K |
| Input $ / M tokens | — | $0.10 |
| Output $ / M tokens | — | $0.32 |
| Results tracked | 27 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.5 Instant leads
GPT-5.5 Instant: 43.7 (#73), Llama-3.3-70B-Instruct: 30.6 (#284)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| SciCode | 48.6% | 26% |
| LMArena Coding | 1433 | 1268 |
| WeirdML | — | 14.4% |
| BigCodeBench Instruct | — | 46.9% |
| LiveBench Coding | — | 36.6% |
| BigCodeBench Complete | — | 57.5% |
Agentic & Tool Use Not comparable
GPT-5.5 Instant: —, Llama-3.3-70B-Instruct: 25.6 (#100)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 31.9% |
| BALROG | — | 23% |
Reasoning GPT-5.5 Instant leads
GPT-5.5 Instant: 24.2 (#149), Llama-3.3-70B-Instruct: 13.7 (#325)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| CritPt | 0% | 0% |
| LMArena Hard Prompts | 1426 | 1257 |
| Epoch Capabilities Index | 142.48 | 127.32 |
| SimpleBench | — | 19.9% |
| Chess Puzzles | 12% | — |
| LiveBench Reasoning | — | 50.8% |
| DTBench | — | 59.5% |
| LiveBench Data Analysis | — | 49.5% |
| LMCA | — | 17.5% |
| ForecastBench | — | 58.6 |
| LiveBench | — | 50.2% |
Math GPT-5.5 Instant leads
GPT-5.5 Instant: 26.1 (#241), Llama-3.3-70B-Instruct: 14.3 (#292)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 68.1% | 5.1% |
| LMArena Math | 1420 | 1268 |
| FrontierMath (Tiers 1-3) | 26.3% | — |
| FrontierMath Tier 4 | 2.4% | — |
| LiveBench Math | — | 42.2% |
| MATH Level 5 | — | 41.6% |
Knowledge GPT-5.5 Instant leads
GPT-5.5 Instant: 48.9 (#68), Llama-3.3-70B-Instruct: 30.5 (#220)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| GPQA Diamond | 82.5% | 47.4% |
| LMArena Expert | 1409 | 1225 |
| Confabulations | — | 22.8% |
| Vectara Hallucination Rate | — | 4.1% |
| MMLU | — | 86.3% |
Multimodal Not comparable
GPT-5.5 Instant: 39.3 (#48), Llama-3.3-70B-Instruct: —
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Vision | 1250 | — |
| LMArena Document | 1403 | — |
Multilingual GPT-5.5 Instant leads
GPT-5.5 Instant: 53.0 (#76), Llama-3.3-70B-Instruct: 40.1 (#215)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Non-English | 1417 | 1236 |
| LMArena Chinese | 1457 | 1217 |
| LMArena French | 1428 | 1280 |
| LMArena German | 1410 | 1251 |
| LMArena Japanese | 1408 | 1149 |
| LMArena Korean | 1393 | 1142 |
| LMArena Russian | 1430 | 1252 |
| LMArena Spanish | 1430 | 1269 |
Instruction Following GPT-5.5 Instant leads
GPT-5.5 Instant: 74.4 (#94), Llama-3.3-70B-Instruct: 71.3 (#152)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Instruction Following | 1407 | 1242 |
| LiveBench Instruction Following | — | 82.7% |
Long Context GPT-5.5 Instant leads
GPT-5.5 Instant: 43.4 (#92), Llama-3.3-70B-Instruct: 26.4 (#290)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Longer Query | 1422 | 1256 |
| Fiction.LiveBench | — | 33.3% |
Writing & Preference GPT-5.5 Instant leads
GPT-5.5 Instant: 62.0 (#81), Llama-3.3-70B-Instruct: 47.8 (#203)
| Benchmark | GPT-5.5 Instant | Llama-3.3-70B-Instruct |
|---|---|---|
| LMArena Text | 1419 | 1274 |
| LMArena Creative Writing | 1420 | 1250 |
| LMArena Multi-Turn | 1434 | 1279 |
| LiveBench Language | — | 39.2% |
Frequently asked questions
Is GPT-5.5 Instant better than Llama-3.3-70B-Instruct?
GPT-5.5 Instant is the stronger model overall, scoring 42.5 to 30.3 on the Noometry Index.
Is GPT-5.5 Instant or Llama-3.3-70B-Instruct better for coding?
GPT-5.5 Instant scores higher on coding benchmarks: 43.7 versus 30.6 in the Noometry coding category.
How many benchmarks do GPT-5.5 Instant and Llama-3.3-70B-Instruct share?
22 benchmarks have published results for both models. GPT-5.5 Instant has 27 scored results on Noometry and Llama-3.3-70B-Instruct has 43.