Claude Fable 5 vs DeepSeek-V4-Pro
Wins 6 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents
Wins 0 of 7 areas
—
Claude Fable 5 is the stronger all-rounder.DeepSeek-V4-Pro is cheaper.
Scores updated · 51 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software90Claude Fable 59 of 9 tests
- AgentsCarrying out multi-step tasks on its own60Claude Fable 56 of 6 tests
- ReasoningHard problems that need careful thinking50Claude Fable 55 of 5 tests
- FactsGetting facts right instead of making them up30Claude Fable 53 of 3 tests
- MathCompetition and research-level math30Claude Fable 53 of 3 tests
- Long documentsFinding answers in very long texts10Claude Fable 51 of 1 test
- Following instructionsDoing exactly what it is asked11Even1 each
Long documents rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
DeepSeek-V4-Pro costs 98% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Claude Fable 5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+87.8points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+41.7points ahead
- Common-sense trick questionsSimpleBench+31points ahead
Where DeepSeek-V4-Pro pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+13points ahead
Every test, side by side
All 51 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Fable 5
- SWE-bench ProClaude Fable 5 by 24.68055.4+24.6
- Terminal-Bench 2.1Claude Fable 5 by 20.684.664+20.6
- LiveBench · Agentic CodingClaude Fable 5 by 19.662.242.6+19.6
- Terminal-Bench HardClaude Fable 5 by 16.762.946.2+16.7
- LiveBench · CodingClaude Fable 5 by 168670+16
- LMArena · WebDevClaude Fable 5 by 160 rating points16241464+160 rating
- SWE-bench VerifiedClaude Fable 5 by 14.49580.6+14.4
- SWE-bench MultilingualClaude Fable 5 by 10.486.676.2+10.4
- SciCodeClaude Fable 5 by 10.26150.8+10.2
AgentsClaude Fable 5
- Terminal-Bench 4.0Claude Fable 5 by 27.842.414.6+27.8
- GDPValClaude Fable 5 by 22.755.632.9+22.7
- AA ApexAgentsClaude Fable 5 by 1943.324.3+19
- MCP AtlasClaude Fable 5 by 11.184.773.6+11.1
- τ-Bench V3 · BankingClaude Fable 5 by 838.130.1+8
- BrowseCompClaude Fable 5 by 487.483.4+4
ReasoningClaude Fable 5
- SimpleBenchClaude Fable 5 by 3181.950.9+31
- Humanity's Last ExamClaude Fable 5 by 1855.537.5+18
- CritPtClaude Fable 5 by 15.728.612.9+15.7
- LiveBench · ReasoningClaude Fable 5 by 789.782.7+7
- GPQA DiamondClaude Fable 5 by 3.892.688.8+3.8
FactsClaude Fable 5
- AA-Omniscience · Non-hallucinationClaude Fable 5 by 30.536.45.9+30.5
- SimpleQA VerifiedClaude Fable 5 by 24.570.746.2+24.5
- AA-Omniscience · AccuracyClaude Fable 5 by 22.465.343+22.4
MathClaude Fable 5
- FrontierMath Tier 4Claude Fable 5 by 87.890.22.4+87.8
- FrontierMath Tiers 1-3 (v2)Claude Fable 5 by 41.78745.3+41.7
- LiveBench · MathematicsClaude Fable 5 by 5.39690.7+5.3
Following instructionsEven
- LiveBench · Instruction FollowingClaude Fable 5 by 13.475.862.4+13.4
- IFBenchDeepSeek-V4-Pro by 1363.576.5+13
Other results22 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- DeepSWEClaude Fable 5 by 57.27012.8+57.2
- AA-OmniscienceClaude Fable 5 by 5443.3-10.7+54
- GDPval-AA v2Claude Fable 5 by 417 rating points17231306+417 rating
- Program BenchClaude Fable 5 by 38.586.347.8+38.5
- GDPval-AA (Elo)Claude Fable 5 by 378 rating points19321554+378 rating
- Terminal-Bench 4.0Claude Fable 5 by 32.144.512.4+32.1
- SWEBench Pro PublicClaude Fable 5 by 24.68055.4+24.6
- AA Agentic IndexClaude Fable 5 by 23.35127.7+23.3
- Toolathlon VerifiedClaude Fable 5 by 2277.955.9+22
- AA IntelligenceClaude Fable 5 by 19.249.630.4+19.2
- Artificial Analysis Coding IndexClaude Fable 5 by 17.176.559.4+17.1
- livebench_languageClaude Fable 5 by 12.690.778.1+12.6
- ToolathlonClaude Fable 5 by 9.961.751.8+9.9
- livebench_data_analysisClaude Fable 5 by 680.574.5+6
- LiveBenchClaude Fable 5 by 4.778.373.6+4.7
- BrowseComp (w/ Ctx)Claude Fable 5 by 4.68883.4+4.6
- Global-MMLU-LiteClaude Fable 5 by 493.389.3+4
- HLE (with tools)Claude Fable 5 by 3.863.860+3.8
- τ²-Bench Telecom (AA run)Claude Fable 5 by 2.398.596.2+2.3
- τ³-Bench Bankingtie26.826tie
- StrongREJECTtie98.798.6tie
- Agents' Last Examtie25.725.7tie
Questions people ask
Which is better, Claude Fable 5 or DeepSeek-V4-Pro?
Claude Fable 5 wins six of the seven areas where both have results: coding, agents, reasoning, facts, math and long documents. DeepSeek-V4-Pro wins none, but costs 98% less. They are level on following instructions.
Which is better for coding?
Claude Fable 5. It wins 9 of the 9 coding tests both models report; DeepSeek-V4-Pro wins none.
Which is cheaper?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; DeepSeek-V4-Pro costs $0.43 and $0.87. That makes DeepSeek-V4-Pro about 98% cheaper for the same work.
How do you compare the two?
We use the 51 benchmark tests both models have published scores on. The verdict counts the 29 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.