Model comparison
Claude Fable 5 vs DeepSeek V4 Flash
Claude Fable 5 is the stronger model overall, scoring 66.8 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 76× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.
Last verified . 40 shared benchmarks.
DeepSeekSummary
- They share 40 benchmarks with published results for both. Claude Fable 5 scores higher in 8 categories and DeepSeek V4 Flash in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Fable 5 leads 88.5 to 60.3.
- The biggest single-benchmark swing is FrontierMath Tier 4: 90.2% for Claude Fable 5 and 24.4% for DeepSeek V4 Flash.
- DeepSeek V4 Flash is cheaper at $0.15 / $0.60 per million input/output tokens, against $10 / $50 for Claude Fable 5.
- DeepSeek V4 Flash has downloadable open weights; the other is API-only.
Side by side
| Claude Fable 5 | DeepSeek V4 Flash | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 66.8 | 53.6 |
| Released | 2026-06-07 | 2026-04-24 |
| Weights | Proprietary | Open |
| Context window | 1M | 1M |
| Max output | 128K | 393K |
| Input $ / M tokens | $10 | $0.15 |
| Output $ / M tokens | $50 | $0.60 |
| Results tracked | 62 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Fable 5 leads
Claude Fable 5: 70.6 (#4), DeepSeek V4 Flash: 47.9 (#59)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| FrontierCode | 53.5% | 18.8% |
| LMArena WebDev | 1625 | 1582 |
| SciCode | 61% | 49.9% |
| WeirdML | 91.9% | 63% |
| LMArena Coding | 1519 | 1457 |
| ALE-Bench | 2,041 | 1,306 |
| DeepSWE | 69.9% | — |
| FrontierSWE | 47% | — |
| GSO | 78.4% | — |
| MirrorCode | 63.9% | — |
Agentic & Tool Use Not comparable
Claude Fable 5: 54.0 (#2), DeepSeek V4 Flash: —
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| APEX-Agents | 63.6% | — |
| Remote Labor Index | 16.1% | — |
| τ²-bench Banking | 39.7% | — |
| PostTrainBench | 41.8% | — |
| GBAEval | 74.5% | — |
| GDP.pdf | 30% | — |
| LMArena Search | 1230 | — |
| Vending-Bench 2 | 5,680 | — |
Reasoning Claude Fable 5 leads
Claude Fable 5: 76.8 (#6), DeepSeek V4 Flash: 53.7 (#30)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| ARC-AGI-2 | 89.2% | 61.4% |
| SimpleBench | 81.9% | 61.1% |
| Kagi LLM Benchmark | 91.4% | 52.2% |
| NYT Connections (extended) | 92.7% | 89.6% |
| ARC-AGI-1 | 98.5% | 89% |
| CritPt | 28.6% | 16.6% |
| Chess Puzzles | 41% | 33% |
| LMArena Hard Prompts | 1508 | 1444 |
| Mystery Game Puzzles | 52% | 34% |
| DTBench | 98.4% | 90.9% |
| LMCA | 61.1% | 41.7% |
| Epoch Capabilities Index | 162.06 | 154.49 |
| EnigmaEval | 39.3% | — |
| EBR-Bench | 39.5% | — |
| Surface Evolver Bench | 95% | — |
| Bench to the Future 3 | 0.13 | — |
Math Claude Fable 5 leads
Claude Fable 5: 88.5 (#5), DeepSeek V4 Flash: 60.3 (#37)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| FrontierMath (Tiers 1-3) | 87% | 57.5% |
| FrontierMath Tier 4 | 90.2% | 24.4% |
| OTIS Mock AIME 2024-2025 | 100% | 94.4% |
| ProofBench | 95% | 56% |
| LMArena Math | 1519 | 1427 |
| MathArena Final-Answer Competitions | — | 76.5% |
| FrontierMath Erdős | 0% | — |
Knowledge Claude Fable 5 leads
Claude Fable 5: 62.2 (#25), DeepSeek V4 Flash: 55.4 (#48)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| GPQA Diamond | 85.9% | 91% |
| SimpleQA Verified | 70.7% | 33.6% |
| LMArena Expert | 1534 | 1441 |
Multimodal Not comparable
Claude Fable 5: 45.3 (#17), DeepSeek V4 Flash: —
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Vision | 1324 | — |
| Blueprint-Bench 2 | 38.6% | — |
| Furniture Assembly | 35.8% | — |
| LMArena Document | 1496 | — |
Multilingual Claude Fable 5 leads
Claude Fable 5: 57.3 (#9), DeepSeek V4 Flash: 53.0 (#72)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Non-English | 1481 | 1420 |
| LMArena Chinese | 1543 | 1468 |
| LMArena French | 1505 | 1439 |
| LMArena German | 1486 | 1418 |
| LMArena Japanese | 1506 | 1406 |
| LMArena Korean | 1488 | 1384 |
| LMArena Russian | 1504 | 1428 |
| LMArena Spanish | 1498 | 1436 |
Instruction Following Claude Fable 5 leads
Claude Fable 5: 78.6 (#8), DeepSeek V4 Flash: 74.9 (#81)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Instruction Following | 1502 | 1421 |
Long Context Claude Fable 5 leads
Claude Fable 5: 46.3 (#23), DeepSeek V4 Flash: 43.8 (#85)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Longer Query | 1509 | 1434 |
Writing & Preference Claude Fable 5 leads
Claude Fable 5: 75.9 (#5), DeepSeek V4 Flash: 63.8 (#61)
| Benchmark | Claude Fable 5 | DeepSeek V4 Flash |
|---|---|---|
| LMArena Text | 1491 | 1432 |
| LMArena Creative Writing | 1494 | 1403 |
| EQ-Bench Creative Writing | 1943 | 1559 |
| LMArena Multi-Turn | 1504 | 1449 |
| EQ-Bench 4 | 1340 | — |
Frequently asked questions
Is Claude Fable 5 better than DeepSeek V4 Flash?
Claude Fable 5 is the stronger model overall, scoring 66.8 to 53.6 on the Noometry Index. DeepSeek V4 Flash costs 76× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.
Which is cheaper, Claude Fable 5 or DeepSeek V4 Flash?
DeepSeek V4 Flash is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; Claude Fable 5 lists at $10 and $50.
Is Claude Fable 5 or DeepSeek V4 Flash better for coding?
Claude Fable 5 scores higher on coding benchmarks: 70.6 versus 47.9 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do Claude Fable 5 and DeepSeek V4 Flash share?
40 benchmarks have published results for both models. Claude Fable 5 has 62 scored results on Noometry and DeepSeek V4 Flash has 41.