Codestral 2508 vs R1 0528
Codestral 2508 costs 2.4x less per output token than R1 0528 ($0.90 vs $2.18 per million). Updated October 21, 2026; 7 benchmarks have a result for both.
Codestral 2508 or R1 0528: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Codestral 2508 for output tokens 2.4x cheaper ($0.90 vs $2.18 per million) and a larger context window (256K against 164K tokens).
- Shared results
- 7
- Codestral 2508 only
- 1
- R1 0528 only
- 53
- Like-for-like categories
- 2 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
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, six public coding benchmarks
No call
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
No call
Only R1 0528 has a AA Intelligence Index result, 13.1.
Directional only
Web design
Design Arena: websites
R1 0528
R1 0528 leads, 1149 against 1013.
Same board, both measured
Long documents
Listed context window
Codestral 2508
Codestral 2508 takes 256K tokens against 164K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Codestral 2508
Codestral 2508 $0.0008, R1 0528 $0.0016 per turn.
List prices
At a glance
- Coding index
- Codestral 2508 not benchmarked, R1 0528 not benchmarked.
- Widest category
- Coding, 22.2 points apart on average over 1 shared benchmark.
- Widest single gap
- ALE-Bench: Codestral 2508 137.8 against R1 0528 804.1, 22.2 points apart on a 0 to 100 scale.
- Output price
- Codestral 2508 $0.90 against R1 0528 $2.18 per million output tokens, 2.4x apart.
- Context window
- Codestral 2508 takes 256K tokens, R1 0528 164K.
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 | Codestral 2508 | Bars | R1 0528 | Ties |
|---|---|---|---|---|
| Coding | 0 | 1 | ||
| Web design | 0 | 6 | ||
| All shared boards | 0 | 7 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index01
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| Aider polyglot | Not measured | 71.4% | Only R1 0528 measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 77.0% | Only R1 0528 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 15.9% | Only R1 0528 measured |
| WeirdML | Not measured | 41.6% | Only R1 0528 measured |
| ALE-Bench | 137.8 | 804.1 | R1 0528 |
| LMArena Text: coding | Not measured | 1427 | Only R1 0528 measured |
| LMArena Text: software and IT | Not measured | 1435 | Only R1 0528 measured |
| Neither model has a result on the other 32 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| AA Intelligence Index | Not measured | 13.1 | Only R1 0528 measured |
| Neither model has a result on the other 1 board here | |||
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| τ²-Bench Telecom | Not measured | 36.5% | Only R1 0528 measured |
| METR time horizons | Not measured | 53.8% | Only R1 0528 measured |
| DeepResearch Bench | Not measured | 35.1% | Only R1 0528 measured |
| BrowseComp-Plus | Not measured | 16.4% | Only R1 0528 measured |
| Neither model has a result on the other 37 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | Not measured | 81.3% | Only R1 0528 measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 15.8% | Only R1 0528 measured |
| GPQA Diamond (Epoch AI run) | Not measured | 76.3% | Only R1 0528 measured |
| ARC-AGI | Not measured | 21.2% | Only R1 0528 measured |
| ARC-AGI-2 | Not measured | 1.1% | Only R1 0528 measured |
| SimpleBench | Not measured | 40.8% | Only R1 0528 measured |
| LMArena Text: hard prompts | Not measured | 1415 | Only R1 0528 measured |
| LMArena Text: expert | Not measured | 1394 | Only R1 0528 measured |
| LMArena Text: hard prompts in English | Not measured | 1422 | Only R1 0528 measured |
| Kagi LLM Benchmark | 32.5% | Not measured | Only Codestral 2508 measured |
| Neither model has a result on the other 12 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| AA Math Index | Not measured | 76.0 | Only R1 0528 measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 76.0% | Only R1 0528 measured |
| OTIS Mock AIME | Not measured | 66.4% | Only R1 0528 measured |
| MATH level 5 | Not measured | 96.6% | Only R1 0528 measured |
| LMArena Text: math | Not measured | 1400 | Only R1 0528 measured |
| LMArena Text: mathematical work | Not measured | 1415 | Only R1 0528 measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index5 boards, none shared
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | Not measured | 84.9% | Only R1 0528 measured |
| LMArena Text: science | Not measured | 1430 | Only R1 0528 measured |
| LMArena Text: medicine and healthcare | Not measured | 1445 | Only R1 0528 measured |
| LMArena Text: legal and government | Not measured | 1419 | Only R1 0528 measured |
| LMArena Text: business and finance | Not measured | 1402 | Only R1 0528 measured |
| Neither model has a result on the other 13 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| Design Arena: websites | 1013 | 1149 | R1 0528 |
| Design Arena: UI components | 1017 | 1103 | R1 0528 |
| Design Arena: game dev | 968 | 1097 | R1 0528 |
| Design Arena: data viz | 1017 | 1182 | R1 0528 |
| Design Arena: 3D | 1020 | 1111 | R1 0528 |
| Design Arena: all code categories | 1007 | 1139 | R1 0528 |
| Design Arena: SVG | Not measured | 1039 | Only R1 0528 measured |
| Neither model has a result on the other 2 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index21 boards, none shared
| Benchmark | Codestral 2508 | R1 0528 | Leader |
|---|---|---|---|
| IFBench | Not measured | 39.6% | Only R1 0528 measured |
| Long context reasoning (AA-LCR) | Not measured | 55.7% | Only R1 0528 measured |
| Fiction.LiveBench | Not measured | 75.0% | Only R1 0528 measured |
| Creative writing (Lech Mazur) | Not measured | 81.9% | Only R1 0528 measured |
| LMArena Text | Not measured | 1428 | Only R1 0528 measured |
| LMArena Text: instruction following | Not measured | 1382 | Only R1 0528 measured |
| LMArena Text: creative writing | Not measured | 1405 | Only R1 0528 measured |
| LMArena Text: English | Not measured | 1437 | Only R1 0528 measured |
| LMArena Text: non-English | Not measured | 1412 | Only R1 0528 measured |
| LMArena Text: Chinese | Not measured | 1442 | Only R1 0528 measured |
| LMArena Text: German | Not measured | 1404 | Only R1 0528 measured |
| LMArena Text: Japanese | Not measured | 1391 | Only R1 0528 measured |
| LMArena Text: Korean | Not measured | 1360 | Only R1 0528 measured |
| LMArena Text: French | Not measured | 1417 | Only R1 0528 measured |
| LMArena Text: Spanish | Not measured | 1411 | Only R1 0528 measured |
| LMArena Text: Russian | Not measured | 1423 | Only R1 0528 measured |
| LMArena Text: Polish | Not measured | 1428 | Only R1 0528 measured |
| LMArena Text: multi-turn | Not measured | 1405 | Only R1 0528 measured |
| LMArena Text: longer queries | Not measured | 1391 | Only R1 0528 measured |
| LMArena Text: writing and language | Not measured | 1402 | Only R1 0528 measured |
| LMArena Text: entertainment and media | Not measured | 1400 | Only R1 0528 measured |
| Neither model has a result on the other 25 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Codestral 2508 | R1 0528 | 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
Codestral 2508 costs less
Repository review
50K input + 3K output tokens
Codestral 2508 costs less
Agent loop
200K input + 10K output tokens
Codestral 2508 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 | Mistral Codestral 2508 | DeepSeek R1 0528 |
|---|---|---|
| Maker | Mistral AI | DeepSeek |
| Weights | No public weights listed | Open weights |
| Released | August 1, 2025 | May 28, 2025 |
| Context window | 256K | 164K |
| Input $/Mtok | $0.30 | $0.50 |
| Output $/Mtok | $0.90 | $2.18 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Codestral 2508 better than R1 0528 for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Codestral 2508 for output tokens 2.4x cheaper ($0.90 vs $2.18 per million) and a larger context window (256K against 164K tokens).
Which is cheaper, Codestral 2508 or R1 0528?
Codestral 2508 costs $0.30 per million input tokens and $0.90 per million output tokens; R1 0528 costs $0.50 and $2.18. Cost per solved task is known for neither of them.
Which has the bigger context window, Codestral 2508 or R1 0528?
Codestral 2508, with 256K tokens against 164K.
How many benchmarks is the Codestral 2508 vs R1 0528 comparison based on?
7 benchmarks have a result for both models as of October 21, 2026. Codestral 2508's coding index rests on 0 independent benchmarks and R1 0528's on 0. Where only one model has a result, the page says so instead of naming a winner.