Claude Haiku 5.5 vs DeepSeek V4 Flash Vision Exp
Updated October 21, 2026. 11 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
Decision reading
Pick Claude Haiku 5.5 for coding: it scores 75.9 on the Vibe Coding Index against 37.6 for DeepSeek V4 Flash Vision Exp.
Pick DeepSeek V4 Flash Vision Exp when the context window is larger, 1.05M against 1M tokens, or when what matters is LiveBench reasoning or LiveBench data analysis, where it leads.
- Shared results
- 11
- Claude Haiku 5.5 only
- 51
- DeepSeek V4 Flash Vision Exp only
- 3
- Like-for-like categories
- 6 of 9
Shared results can support a head-to-head reading. Results only one model has describe coverage, not superiority.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, four public coding benchmarks
Claude Haiku 5.5
Claude Haiku 5.5 scores 75.9 against 37.6.
2 and 1 benchmarks behind the scores
Intelligence
AA Intelligence Index
Claude Haiku 5.5
Claude Haiku 5.5 leads, 43.4 against 34.8.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only DeepSeek V4 Flash Vision Exp has a AA Agentic Index result, 47.5.
Directional only
Web design
Design Arena: websites
No call
Only Claude Haiku 5.5 has a Design Arena: websites result, 1263.
Directional only
General
Vals Index
No call
Only Claude Haiku 5.5 has a Vals Index result, 58.7%.
Directional only
Long documents
Listed context window
DeepSeek V4 Flash Vision Exp
DeepSeek V4 Flash Vision Exp takes 1.05M tokens against 1M.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Claude Haiku 5.5
Claude Haiku 5.5 $0.0003, DeepSeek V4 Flash Vision Exp $0.0005 per turn.
List prices
At a glance
- Coding index
- Claude Haiku 5.5 75.9 from 2 benchmarks, DeepSeek V4 Flash Vision Exp 37.6 from 1 benchmark.
- Widest category
- Coding, 29.6 points apart on average over 2 shared benchmarks.
- Widest single gap
- LiveBench coding: Claude Haiku 5.5 91.4% against DeepSeek V4 Flash Vision Exp 37.6%, 53.8 points apart on a 0 to 100 scale.
- Output price
- Claude Haiku 5.5 $0.50 against DeepSeek V4 Flash Vision Exp $0.6468 per million output tokens, 1.3x apart.
- Context window
- Claude Haiku 5.5 takes 1M tokens, DeepSeek V4 Flash Vision Exp 1.05M.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Claude Haiku 5.5 | Bars | DeepSeek V4 Flash Vision Exp | Ties |
|---|---|---|---|---|
| Coding | 2 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 3 | 0 | ||
| Reasoning | 1 | 1 | ||
| Web design | 1 | 0 | ||
| General | 1 | 1 | ||
| All shared boards | 9 | 2 |
CodingLike-for-likeFour of these boards feed the Vibe Coding Index20
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| LiveBench coding | 91.4% | 37.6% | Claude Haiku 5.5 |
| LMArena WebDev | 1586 | Not measured | Only Claude Haiku 5.5 measured |
| AA Coding Index | Not measured | 65.0 | Only DeepSeek V4 Flash Vision Exp measured |
| Terminal-Bench 4 (Vals AI) | 36.4% | Not measured | Only Claude Haiku 5.5 measured |
| Vibe Code Bench (Vals AI) | 90.4% | Not measured | Only Claude Haiku 5.5 measured |
| Code migration (Vals AI) | 44.6% | Not measured | Only Claude Haiku 5.5 measured |
| ProgramBench (Vals AI) | 2.5% | Not measured | Only Claude Haiku 5.5 measured |
| IOI (Vals AI) | 47.3% | Not measured | Only Claude Haiku 5.5 measured |
| SWE-bench Multimodal (Vals AI) | 44.0% | Not measured | Only Claude Haiku 5.5 measured |
| Vibe Code Bench 1-100 (Vals AI) | 20.4% | Not measured | Only Claude Haiku 5.5 measured |
| CyberBench (Vals AI) | 75.5% | Not measured | Only Claude Haiku 5.5 measured |
| SciCode | 55.0% | 49.7% | Claude Haiku 5.5 |
| FrontierCode | 46.4% | Not measured | Only Claude Haiku 5.5 measured |
| FrontierSWE | 43.8% | Not measured | Only Claude Haiku 5.5 measured |
| LMArena WebDev: React | 1605 | Not measured | Only Claude Haiku 5.5 measured |
| LMArena WebDev: HTML | 1569 | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 22 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| AA Intelligence Index | 43.4 | 34.8 | Claude Haiku 5.5 |
| Epoch Capabilities Index | 159.3 | Not measured | Only Claude Haiku 5.5 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| AA Agentic Index | Not measured | 47.5 | Only DeepSeek V4 Flash Vision Exp measured |
| LiveBench research | 81.9% | 69.2% | Claude Haiku 5.5 |
| LiveBench browser | 80.0% | 64.0% | Claude Haiku 5.5 |
| LiveBench long agentic | 57.6% | 39.0% | Claude Haiku 5.5 |
| SREBench (Vals AI) | 5.7% | Not measured | Only Claude Haiku 5.5 measured |
| Finance Agent (Vals AI) | 54.1% | Not measured | Only Claude Haiku 5.5 measured |
| Tax Agent Bench (Vals AI) | 31.9% | Not measured | Only Claude Haiku 5.5 measured |
| Legal Research Bench (Vals AI) | 43.3% | Not measured | Only Claude Haiku 5.5 measured |
| Harvey's Legal Agent Benchmark (Vals AI) | 4.2% | Not measured | Only Claude Haiku 5.5 measured |
| Excel Modeling Benchmark (Vals AI) | 59.0% | Not measured | Only Claude Haiku 5.5 measured |
| BioMysteryBench (Vals AI) | 70.4% | Not measured | Only Claude Haiku 5.5 measured |
| Terminal-Bench-Science (Vals AI) | 10.0% | Not measured | Only Claude Haiku 5.5 measured |
| Vals Science Index | 38.5% | Not measured | Only Claude Haiku 5.5 measured |
| APEX-Agents | 56.1% | Not measured | Only Claude Haiku 5.5 measured |
| Vending-Bench 2 | $3,602 | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 24 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| LiveBench reasoning | 82.4% | 85.4% | DeepSeek V4 Flash Vision Exp |
| MysteryMechanism (Vals AI) | 35.1% | Not measured | Only Claude Haiku 5.5 measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 91.3% | Only DeepSeek V4 Flash Vision Exp measured |
| Humanity's Last Exam (Artificial Analysis run) | 44.4% | 34.5% | Claude Haiku 5.5 |
| GPQA Diamond (Epoch AI run) | 89.6% | Not measured | Only Claude Haiku 5.5 measured |
| ARC-AGI | 89.0% | Not measured | Only Claude Haiku 5.5 measured |
| ARC-AGI-2 | 61.3% | Not measured | Only Claude Haiku 5.5 measured |
| Chess puzzles | 55.0% | Not measured | Only Claude Haiku 5.5 measured |
| Mystery game puzzles | 30.0% | Not measured | Only Claude Haiku 5.5 measured |
| EBR-bench | 43.8% | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 11 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| OTIS Mock AIME | 98.9% | Not measured | Only Claude Haiku 5.5 measured |
| FrontierMath, tiers 1 to 3 | 75.1% | Not measured | Only Claude Haiku 5.5 measured |
| FrontierMath, tier 4 | 46.3% | Not measured | Only Claude Haiku 5.5 measured |
| ProofBench | 89.0% | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 9 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| MedCode (Vals AI) | 51.5% | Not measured | Only Claude Haiku 5.5 measured |
| SimpleQA Verified | 23.8% | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 15 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| Design Arena: websites | 1263 | Not measured | Only Claude Haiku 5.5 measured |
| Design Arena: UI components | 1295 | Not measured | Only Claude Haiku 5.5 measured |
| Design Arena: game dev | 1315 | Not measured | Only Claude Haiku 5.5 measured |
| Design Arena: data viz | 1292 | Not measured | Only Claude Haiku 5.5 measured |
| Design Arena: 3D | 1343 | Not measured | Only Claude Haiku 5.5 measured |
| Design Arena: all code categories | 1279 | Not measured | Only Claude Haiku 5.5 measured |
| LiveBench design | 83.1% | 70.7% | Claude Haiku 5.5 |
| Neither model has a result on the other 2 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| Vals Index | 58.7% | Not measured | Only Claude Haiku 5.5 measured |
| LiveBench data analysis | 53.1% | 79.5% | DeepSeek V4 Flash Vision Exp |
| Public Benefits Bench (Vals AI) | 63.2% | Not measured | Only Claude Haiku 5.5 measured |
| SAGE (Vals AI) | 45.8% | Not measured | Only Claude Haiku 5.5 measured |
| MedScribe (Vals AI) | 89.5% | Not measured | Only Claude Haiku 5.5 measured |
| Long context reasoning (AA-LCR) | 82.7% | 81.3% | Claude Haiku 5.5 |
| DTBench | 95.0% | Not measured | Only Claude Haiku 5.5 measured |
| LMCA | 51.1% | Not measured | Only Claude Haiku 5.5 measured |
| Furniture assembly | 47.5% | Not measured | Only Claude Haiku 5.5 measured |
| Neither model has a result on the other 37 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Claude Haiku 5.5 costs less
Repository review
50K input + 3K output tokens
Claude Haiku 5.5 costs less
Agent loop
200K input + 10K output tokens
Claude Haiku 5.5 costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Anthropic Claude Haiku 5.5 | DeepSeek V4 Flash Vision Exp |
|---|---|---|
| Maker | Anthropic | DeepSeek |
| Weights | No public weights listed | Open weights |
| Released | October 7, 2026 | August 21, 2026 |
| Context window | 1M | 1.05M |
| Input $/Mtok | $0.10 | $0.2156 |
| Output $/Mtok | $0.50 | $0.6468 |
| Coding index | 75.9 | 37.6 |
| Independent measurements behind it | 2 | 1 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Haiku 5.5 better than DeepSeek V4 Flash Vision Exp for coding?
Pick Claude Haiku 5.5 for coding: it scores 75.9 on the Vibe Coding Index against 37.6 for DeepSeek V4 Flash Vision Exp. Pick DeepSeek V4 Flash Vision Exp when the context window is larger, 1.05M against 1M tokens, or when what matters is LiveBench reasoning or LiveBench data analysis, where it leads.
Which is cheaper, Claude Haiku 5.5 or DeepSeek V4 Flash Vision Exp?
Claude Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens; DeepSeek V4 Flash Vision Exp costs $0.2156 and $0.6468. Cost per solved task is known for neither of them.
Which has the bigger context window, Claude Haiku 5.5 or DeepSeek V4 Flash Vision Exp?
DeepSeek V4 Flash Vision Exp, with 1.05M tokens against 1M.
How many benchmarks is the Claude Haiku 5.5 vs DeepSeek V4 Flash Vision Exp comparison based on?
11 benchmarks have a result for both models as of October 21, 2026. Claude Haiku 5.5's coding index rests on 2 independent benchmarks and DeepSeek V4 Flash Vision Exp's on 1. Where only one model has a result, the page says so instead of naming a winner.