Model comparison
DeepSeek V4 Flash vs Trinity Large Thinking
DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 38.6 on the Noometry Index.
Last verified . 21 shared benchmarks.
DeepSeek
Arcee AISummary
- They share 21 benchmarks with published results for both. DeepSeek V4 Flash scores higher in 8 categories and Trinity Large Thinking in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Flash leads 53.7 to 16.9.
- The biggest single-benchmark swing is NYT Connections (extended): 89.6% for DeepSeek V4 Flash and 16.5% for Trinity Large Thinking.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.25 / $0.80 for Trinity Large Thinking.
- DeepSeek V4 Flash accepts more context: 1M tokens versus 262K.
Side by side
| DeepSeek V4 Flash | Trinity Large Thinking | |
|---|---|---|
| Provider | DeepSeek | Arcee AI |
| Noometry Index | 53.6 | 38.6 |
| Released | 2026-04-24 | 2026-04-01 |
| Weights | Open | Open |
| Context window | 1M | 262K |
| Max output | 393K | 80K |
| Input $ / M tokens | $0.15 | $0.25 |
| Output $ / M tokens | $0.60 | $0.80 |
| Results tracked | 41 | 24 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Flash leads
DeepSeek V4 Flash: 47.9 (#59), Trinity Large Thinking: 34.1 (#244)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena WebDev | 1582 | 1238 |
| SciCode | 49.9% | 36.1% |
| LMArena Coding | 1457 | 1381 |
| FrontierCode | 18.8% | — |
| WeirdML | 63% | — |
| ALE-Bench | 1,306 | — |
Reasoning DeepSeek V4 Flash leads
DeepSeek V4 Flash: 53.7 (#30), Trinity Large Thinking: 16.9 (#298)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| NYT Connections (extended) | 89.6% | 16.5% |
| CritPt | 16.6% | 0.9% |
| LMArena Hard Prompts | 1444 | 1350 |
| ARC-AGI-2 | 61.4% | — |
| SimpleBench | 61.1% | — |
| Kagi LLM Benchmark | 52.2% | — |
| ARC-AGI-1 | 89% | — |
| Chess Puzzles | 33% | — |
| Thematic Generalization | — | 41.6% |
| Mystery Game Puzzles | 34% | — |
| DTBench | 90.9% | — |
| LMCA | 41.7% | — |
| Surface Evolver Bench | — | 15.6% |
| Epoch Capabilities Index | 154.49 | — |
Math DeepSeek V4 Flash leads
DeepSeek V4 Flash: 60.3 (#37), Trinity Large Thinking: 37.6 (#149)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Math | 1427 | 1366 |
| FrontierMath (Tiers 1-3) | 57.5% | — |
| FrontierMath Tier 4 | 24.4% | — |
| MathArena Final-Answer Competitions | 76.5% | — |
| OTIS Mock AIME 2024-2025 | 94.4% | — |
| ProofBench | 56% | — |
Knowledge DeepSeek V4 Flash leads
DeepSeek V4 Flash: 55.4 (#48), Trinity Large Thinking: 40.9 (#113)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Expert | 1441 | 1360 |
| GPQA Diamond | 91% | — |
| SimpleQA Verified | 33.6% | — |
| Vectara Hallucination Rate | — | 6.9% |
Multilingual DeepSeek V4 Flash leads
DeepSeek V4 Flash: 53.0 (#72), Trinity Large Thinking: 46.2 (#160)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Non-English | 1420 | 1325 |
| LMArena Chinese | 1468 | 1373 |
| LMArena French | 1439 | 1374 |
| LMArena German | 1418 | 1356 |
| LMArena Japanese | 1406 | 1311 |
| LMArena Korean | 1384 | 1306 |
| LMArena Russian | 1428 | 1337 |
| LMArena Spanish | 1436 | 1357 |
Instruction Following DeepSeek V4 Flash leads
DeepSeek V4 Flash: 74.9 (#81), Trinity Large Thinking: 70.5 (#162)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Instruction Following | 1421 | 1334 |
Long Context DeepSeek V4 Flash leads
DeepSeek V4 Flash: 43.8 (#85), Trinity Large Thinking: 41.3 (#144)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Longer Query | 1434 | 1355 |
Writing & Preference DeepSeek V4 Flash leads
DeepSeek V4 Flash: 63.8 (#61), Trinity Large Thinking: 53.8 (#158)
| Benchmark | DeepSeek V4 Flash | Trinity Large Thinking |
|---|---|---|
| LMArena Text | 1432 | 1340 |
| LMArena Creative Writing | 1403 | 1320 |
| LMArena Multi-Turn | 1449 | 1342 |
| EQ-Bench Creative Writing | 1559 | — |
Frequently asked questions
Is DeepSeek V4 Flash better than Trinity Large Thinking?
DeepSeek V4 Flash is the stronger model overall, scoring 53.6 to 38.6 on the Noometry Index.
Which is cheaper, DeepSeek V4 Flash or Trinity Large Thinking?
DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Trinity Large Thinking lists at $0.25 and $0.80.
Is DeepSeek V4 Flash or Trinity Large Thinking better for coding?
DeepSeek V4 Flash scores higher on coding benchmarks: 47.9 versus 34.1 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Flash does, with 1M tokens against 262K.
How many benchmarks do DeepSeek V4 Flash and Trinity Large Thinking share?
21 benchmarks have published results for both models. DeepSeek V4 Flash has 41 scored results on Noometry and Trinity Large Thinking has 24.