Codestral 2508 vs Mercury 2
Mercury 2 costs 1.2x less per output token than Codestral 2508 ($0.75 vs $0.90 per million). Updated October 21, 2026; 7 benchmarks have a result for both.
Codestral 2508 or Mercury 2: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick Codestral 2508 for a larger context window (256K against 128K tokens) and work measured by ALE-Bench or Design Arena: websites, where it leads.
- Shared results
- 7
- Codestral 2508 only
- 1
- Mercury 2 only
- 35
- 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 Mercury 2 has a AA Intelligence Index result, 10.8.
Directional only
Agentic work
AA Agentic Index
No call
Only Mercury 2 has a AA Agentic Index result, 1.6.
Directional only
Web design
Design Arena: websites
Mercury 2
Mercury 2 leads, 1036 against 1013.
Same board, both measured
Long documents
Listed context window
Codestral 2508
Codestral 2508 takes 256K tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Mercury 2
Codestral 2508 $0.0008, Mercury 2 $0.0006 per turn.
List prices
At a glance
- Coding index
- Codestral 2508 not benchmarked, Mercury 2 not benchmarked.
- Widest category
- Coding, 21.6 points apart on average over 1 shared benchmark.
- Widest single gap
- ALE-Bench: Codestral 2508 137.8 against Mercury 2 785.6, 21.6 points apart on a 0 to 100 scale.
- Output price
- Codestral 2508 $0.90 against Mercury 2 $0.75 per million output tokens, 1.2x apart.
- Context window
- Codestral 2508 takes 256K tokens, Mercury 2 128K.
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 | Mercury 2 | Ties |
|---|---|---|---|---|
| Coding | 0 | 1 | ||
| Web design | 3 | 3 | ||
| All shared boards | 3 | 4 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index01
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| LMArena WebDev | Not measured | 1171 | Only Mercury 2 measured |
| AA Coding Index | Not measured | 31.1 | Only Mercury 2 measured |
| SciCode | Not measured | 37.7% | Only Mercury 2 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 26.5% | Only Mercury 2 measured |
| WeirdML | Not measured | 43.2% | Only Mercury 2 measured |
| ALE-Bench | 137.8 | 785.6 | Mercury 2 |
| LMArena Text: coding | Not measured | 1390 | Only Mercury 2 measured |
| LMArena Text: software and IT | Not measured | 1385 | Only Mercury 2 measured |
| LMArena WebDev: React | Not measured | 1160 | Only Mercury 2 measured |
| LMArena WebDev: HTML | Not measured | 1199 | Only Mercury 2 measured |
| Neither model has a result on the other 29 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index1 board, none shared
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| AA Intelligence Index | Not measured | 10.8 | Only Mercury 2 measured |
| Neither model has a result on the other 1 board here | |||
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| AA Agentic Index | Not measured | 1.6 | Only Mercury 2 measured |
| τ²-Bench Telecom | Not measured | 70.8% | Only Mercury 2 measured |
| Neither model has a result on the other 39 boards here | |||
ReasoningDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| GPQA Diamond (Artificial Analysis run) | Not measured | 77.0% | Only Mercury 2 measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 17.1% | Only Mercury 2 measured |
| LMArena Text: hard prompts | Not measured | 1362 | Only Mercury 2 measured |
| LMArena Text: expert | Not measured | 1358 | Only Mercury 2 measured |
| LMArena Text: hard prompts in English | Not measured | 1374 | Only Mercury 2 measured |
| Kagi LLM Benchmark | 32.5% | Not measured | Only Codestral 2508 measured |
| Neither model has a result on the other 16 boards here | |||
MathNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| Neither model has a result on 13 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| LMArena Text: science | Not measured | 1343 | Only Mercury 2 measured |
| LMArena Text: medicine and healthcare | Not measured | 1363 | Only Mercury 2 measured |
| LMArena Text: business and finance | Not measured | 1349 | Only Mercury 2 measured |
| Neither model has a result on the other 15 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index33
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| Design Arena: websites | 1013 | 1036 | Mercury 2 |
| Design Arena: UI components | 1017 | 998 | Codestral 2508 |
| Design Arena: game dev | 968 | 985 | Mercury 2 |
| Design Arena: data viz | 1017 | 1009 | Codestral 2508 |
| Design Arena: 3D | 1020 | 983 | Codestral 2508 |
| Design Arena: all code categories | 1007 | 1022 | Mercury 2 |
| Design Arena: SVG | Not measured | 998 | Only Mercury 2 measured |
| Design Arena: ASCII art | Not measured | 988 | Only Mercury 2 measured |
| Neither model has a result on the other 1 board here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index13 boards, none shared
| Benchmark | Codestral 2508 | Mercury 2 | Leader |
|---|---|---|---|
| IFBench | Not measured | 69.8% | Only Mercury 2 measured |
| Long context reasoning (AA-LCR) | Not measured | 43.7% | Only Mercury 2 measured |
| LMArena Text | Not measured | 1354 | Only Mercury 2 measured |
| LMArena Text: instruction following | Not measured | 1328 | Only Mercury 2 measured |
| LMArena Text: creative writing | Not measured | 1286 | Only Mercury 2 measured |
| LMArena Text: English | Not measured | 1378 | Only Mercury 2 measured |
| LMArena Text: non-English | Not measured | 1330 | Only Mercury 2 measured |
| LMArena Text: Chinese | Not measured | 1414 | Only Mercury 2 measured |
| LMArena Text: Russian | Not measured | 1304 | Only Mercury 2 measured |
| LMArena Text: multi-turn | Not measured | 1357 | Only Mercury 2 measured |
| LMArena Text: longer queries | Not measured | 1329 | Only Mercury 2 measured |
| LMArena Text: writing and language | Not measured | 1300 | Only Mercury 2 measured |
| LMArena Text: entertainment and media | Not measured | 1314 | Only Mercury 2 measured |
| Neither model has a result on the other 33 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Codestral 2508 | Mercury 2 | 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
Mercury 2 costs less
Repository review
50K input + 3K output tokens
Mercury 2 costs less
Agent loop
200K input + 10K output tokens
Mercury 2 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 | Inception Mercury 2 |
|---|---|---|
| Maker | Mistral AI | Inception |
| Weights | No public weights listed | No public weights listed |
| Released | August 1, 2025 | March 4, 2026 |
| Context window | 256K | 128K |
| Input $/Mtok | $0.30 | $0.25 |
| Output $/Mtok | $0.90 | $0.75 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 1 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Codestral 2508 better than Mercury 2 for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick Codestral 2508 for a larger context window (256K against 128K tokens) and work measured by ALE-Bench or Design Arena: websites, where it leads.
Which is cheaper, Codestral 2508 or Mercury 2?
Codestral 2508 costs $0.30 per million input tokens and $0.90 per million output tokens; Mercury 2 costs $0.25 and $0.75. Cost per solved task is known for neither of them.
Which has the bigger context window, Codestral 2508 or Mercury 2?
Codestral 2508, with 256K tokens against 128K.
How many benchmarks is the Codestral 2508 vs Mercury 2 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 Mercury 2's on 1. Where only one model has a result, the page says so instead of naming a winner.