Model comparison
Llama-3.3-70B-Instruct vs Trinity Large Thinking
Trinity Large Thinking is the stronger model overall, scoring 38.6 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 2.5× less per token, which makes it the better buy when Trinity Large Thinking's lead doesn't matter for your workload.
Last verified . 20 shared benchmarks.
Meta
Arcee AISummary
- They share 20 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 1 category and Trinity Large Thinking in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Trinity Large Thinking leads 37.6 to 15.3.
- The biggest single-benchmark swing is SciCode: 26% for Llama-3.3-70B-Instruct and 36.1% for Trinity Large Thinking.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $0.25 / $0.80 for Trinity Large Thinking.
- Trinity Large Thinking accepts more context: 262K tokens versus 128K.
Side by side
| Llama-3.3-70B-Instruct | Trinity Large Thinking | |
|---|---|---|
| Provider | Meta | Arcee AI |
| Noometry Index | 30.6 | 38.6 |
| Released | 2024-12-06 | 2026-04-01 |
| Weights | Open | Open |
| Context window | 128K | 262K |
| Max output | 4K | 80K |
| Input $ / M tokens | $0.10 | $0.25 |
| Output $ / M tokens | $0.32 | $0.80 |
| Results tracked | 43 | 24 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 31.0 (#290), Trinity Large Thinking: 34.1 (#244)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| SciCode | 26% | 36.1% |
| LMArena Coding | 1268 | 1381 |
| LMArena WebDev | — | 1238 |
| WeirdML | 14.4% | — |
| BigCodeBench Instruct | 46.9% | — |
| LiveBench Coding | 36.6% | — |
| BigCodeBench Complete | 57.5% | — |
Agentic & Tool Use Not comparable
Llama-3.3-70B-Instruct: 25.8 (#105), Trinity Large Thinking: —
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| Berkeley Function Calling Leaderboard | 31.9% | — |
| BALROG | 23% | — |
Reasoning Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 14.1 (#327), Trinity Large Thinking: 16.9 (#298)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| CritPt | 0% | 0.9% |
| LMArena Hard Prompts | 1257 | 1350 |
| SimpleBench | 19.9% | — |
| NYT Connections (extended) | — | 16.5% |
| Thematic Generalization | — | 41.6% |
| LiveBench Reasoning | 50.8% | — |
| DTBench | 59.5% | — |
| LiveBench Data Analysis | 49.5% | — |
| LMCA | 17.5% | — |
| Surface Evolver Bench | — | 15.6% |
| Epoch Capabilities Index | 127.33 | — |
| ForecastBench | 58.6 | — |
| LiveBench | 50.2% | — |
Math Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 15.3 (#298), Trinity Large Thinking: 37.6 (#149)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| LMArena Math | 1267 | 1366 |
| OTIS Mock AIME 2024-2025 | 5.1% | — |
| LiveBench Math | 42.2% | — |
| MATH Level 5 | 41.6% | — |
Knowledge Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 30.6 (#226), Trinity Large Thinking: 40.9 (#113)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| Vectara Hallucination Rate | 4.1% | 6.9% |
| LMArena Expert | 1225 | 1360 |
| GPQA Diamond | 47.4% | — |
| Confabulations | 22.8% | — |
| MMLU | 86.3% | — |
Multilingual Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 39.9 (#220), Trinity Large Thinking: 46.2 (#160)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| LMArena Non-English | 1236 | 1325 |
| LMArena Chinese | 1217 | 1373 |
| LMArena French | 1281 | 1374 |
| LMArena German | 1251 | 1356 |
| LMArena Japanese | 1150 | 1311 |
| LMArena Korean | 1143 | 1306 |
| LMArena Russian | 1252 | 1337 |
| LMArena Spanish | 1270 | 1357 |
Instruction Following Too close to call
Llama-3.3-70B-Instruct: 71.1 (#157), Trinity Large Thinking: 70.5 (#162)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| LMArena Instruction Following | 1242 | 1334 |
| LiveBench Instruction Following | 82.7% | — |
Long Context Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 26.4 (#295), Trinity Large Thinking: 41.3 (#144)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| LMArena Longer Query | 1256 | 1355 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference Trinity Large Thinking leads
Llama-3.3-70B-Instruct: 47.6 (#207), Trinity Large Thinking: 53.8 (#158)
| Benchmark | Llama-3.3-70B-Instruct | Trinity Large Thinking |
|---|---|---|
| LMArena Text | 1274 | 1340 |
| LMArena Creative Writing | 1250 | 1320 |
| LMArena Multi-Turn | 1280 | 1342 |
| LiveBench Language | 39.2% | — |
Frequently asked questions
Is Llama-3.3-70B-Instruct better than Trinity Large Thinking?
Trinity Large Thinking is the stronger model overall, scoring 38.6 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 2.5× less per token, which makes it the better buy when Trinity Large Thinking's lead doesn't matter for your workload.
Which is cheaper, Llama-3.3-70B-Instruct or Trinity Large Thinking?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Trinity Large Thinking lists at $0.25 and $0.80.
Is Llama-3.3-70B-Instruct or Trinity Large Thinking better for coding?
Trinity Large Thinking scores higher on coding benchmarks: 34.1 versus 31.0 in the Noometry coding category.
Which has the bigger context window?
Trinity Large Thinking does, with 262K tokens against 128K.
How many benchmarks do Llama-3.3-70B-Instruct and Trinity Large Thinking share?
20 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and Trinity Large Thinking has 24.