LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

GPT-4o-mini (2024-07-18) vs Hy3 preview

Updated October 21, 2026. 24 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace GPT-4o-mini (2024-07-18) Replace Hy3 preview
GPT-4o-mini (2024-07-18)OpenAI
Coding
3.6
Overall
No data
Output $/M
$0.60
0
Hy3 previewTencent
Coding
22.4
Overall
25.3
Output $/M
$0.60
24
vs boards won of 24 shared

Decision reading

Pick Hy3 preview for coding: it scores 22.4 on the Vibe Coding Index against 3.6 for GPT-4o-mini (2024-07-18).

GPT-4o-mini (2024-07-18) is neither cheaper nor larger in context, and leads no benchmark both models were measured on.

Model AGPT-4o-mini (2024-07-18)OpenAI3.6/1001 independent benchmarkNo cost per solved taskModel BHy3 previewTencent22.4/1001 independent benchmarkNo cost per solved task
Shared results
24
GPT-4o-mini (2024-07-18) only
22
Hy3 preview only
12
Like-for-like categories
5 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.

At a glance

Coding index
GPT-4o-mini (2024-07-18) 3.6 from 1 benchmark, Hy3 preview 22.4 from 1 benchmark.
Widest category
Math, 27.2 points apart on average over 2 shared benchmarks.
Widest single gap
LMArena Text: expert: GPT-4o-mini (2024-07-18) 1234 against Hy3 preview 1422, 31.2 points apart on a 0 to 100 scale.
Output price
Both cost $0.60 per million output tokens.
Context window
GPT-4o-mini (2024-07-18) takes 128K tokens, Hy3 preview 262K.

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.

Boards won, where both have a result
CategoryGPT-4o-mini (2024-07-18)BarsHy3 previewTies
Coding02
Reasoning03
Math02
Knowledge04
General013
All shared boards024
CodingLike-for-likeFour of these boards feed the Vibe Coding Index02
Coding benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
Aider polyglot3.6%Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena WebDevNot measured1357Only Hy3 preview measured
AA Coding IndexNot measured58.8Only Hy3 preview measured
Terminal-Bench Hard (Artificial Analysis run)Not measured34.1%Only Hy3 preview measured
WeirdML11.8%Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: coding12901421Hy3 preview
LMArena Text: software and IT12911423Hy3 preview
LMArena WebDev: ReactNot measured1349Only Hy3 preview measured
LMArena WebDev: HTMLNot measured1356Only Hy3 preview 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
Intelligence benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
AA Intelligence IndexNot measured25.3Only Hy3 preview measured
Neither model has a result on the other 1 board here
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index3 boards, none shared
Agentic work benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
AA Agentic IndexNot measured24.1Only Hy3 preview measured
τ²-Bench TelecomNot measured92.7%Only Hy3 preview measured
BALROG17.4%Not measuredOnly GPT-4o-mini (2024-07-18) measured
Neither model has a result on the other 36 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index03
Reasoning benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
GPQA Diamond (Artificial Analysis run)Not measured86.7%Only Hy3 preview measured
Humanity's Last Exam (Artificial Analysis run)Not measured27.8%Only Hy3 preview measured
GPQA Diamond (Epoch AI run)37.7%Not measuredOnly GPT-4o-mini (2024-07-18) measured
SimpleBench10.7%Not measuredOnly GPT-4o-mini (2024-07-18) measured
Chess puzzles3.0%Not measuredOnly GPT-4o-mini (2024-07-18) measured
Mystery game puzzles12.0%Not measuredOnly GPT-4o-mini (2024-07-18) measured
VPCT34.0%Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: hard prompts12671413Hy3 preview
LMArena Text: expert12341422Hy3 preview
LMArena Text: hard prompts in English12781416Hy3 preview
Neither model has a result on the other 11 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
OTIS Mock AIME6.9%Not measuredOnly GPT-4o-mini (2024-07-18) measured
FrontierMath, tiers 1 to 30.7%Not measuredOnly GPT-4o-mini (2024-07-18) measured
MATH level 552.6%Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: math12671424Hy3 preview
LMArena Text: mathematical work12711442Hy3 preview
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index04
Knowledge benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
SimpleQA Verified8.3%Not measuredOnly GPT-4o-mini (2024-07-18) measured
MMLU76.7%Not measuredOnly GPT-4o-mini (2024-07-18) measured
GeoBench64.0%Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: science12861416Hy3 preview
LMArena Text: medicine and healthcare12821405Hy3 preview
LMArena Text: legal and government13051430Hy3 preview
LMArena Text: business and finance12651397Hy3 preview
Neither model has a result on the other 10 boards here
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Web design benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
Neither model has a result on 9 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index013
General benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
IFBenchNot measured63.1%Only Hy3 preview measured
Long context reasoning (AA-LCR)Not measured64.7%Only Hy3 preview measured
Creative writing (Lech Mazur)67.2%Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text12861401Hy3 preview
LMArena Text: instruction following12581383Hy3 preview
LMArena Text: creative writing12681339Hy3 preview
LMArena Vision1066Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: English13011409Hy3 preview
LMArena Text: non-English12661390Hy3 preview
LMArena Text: Chinese12651435Hy3 preview
LMArena Text: German1272Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: Japanese1216Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: Korean1195Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: French12971425Hy3 preview
LMArena Text: Spanish12761399Hy3 preview
LMArena Text: Russian12751399Hy3 preview
LMArena Text: Polish1244Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Text: multi-turn12851411Hy3 preview
LMArena Text: longer queries12891402Hy3 preview
LMArena Text: writing and language12821370Hy3 preview
LMArena Text: entertainment and media12531352Hy3 preview
LMArena Vision: English1072Not measuredOnly GPT-4o-mini (2024-07-18) measured
LMArena Vision: Chinese1033Not measuredOnly GPT-4o-mini (2024-07-18) measured
Neither model has a result on the other 23 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for GPT-4o-mini (2024-07-18) and Hy3 preview
BenchmarkGPT-4o-mini (2024-07-18)Hy3 previewLeader
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

GPT-4o-mini (2024-07-18)$0.0004Fits in one request
Hy3 preview$0.0005Fits in one request

GPT-4o-mini (2024-07-18) costs less

Repository review

50K input + 3K output tokens

GPT-4o-mini (2024-07-18)$0.0093Fits in one request
Hy3 preview$0.011Fits in one request

GPT-4o-mini (2024-07-18) costs less

Agent loop

200K input + 10K output tokens

GPT-4o-mini (2024-07-18)$0.036Exceeds the context window
Hy3 preview$0.042Fits in one request

GPT-4o-mini (2024-07-18) 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.

Specifications of GPT-4o-mini (2024-07-18) and Hy3 preview
SpecificationOpenAI GPT-4o-mini (2024-07-18)Tencent Hy3 preview
MakerOpenAITencent
WeightsNo public weights listedOpen weights
ReleasedJuly 18, 2024April 22, 2026
Context window128K262K
Input $/Mtok$0.15$0.18
Output $/Mtok$0.60$0.60
Coding index3.622.4
Independent measurements behind it11
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is GPT-4o-mini (2024-07-18) better than Hy3 preview for coding?

Pick Hy3 preview for coding: it scores 22.4 on the Vibe Coding Index against 3.6 for GPT-4o-mini (2024-07-18). GPT-4o-mini (2024-07-18) is neither cheaper nor larger in context, and leads no benchmark both models were measured on.

Which is cheaper, GPT-4o-mini (2024-07-18) or Hy3 preview?

GPT-4o-mini (2024-07-18) costs $0.15 per million input tokens and $0.60 per million output tokens; Hy3 preview costs $0.18 and $0.60. Cost per solved task is known for neither of them.

Which has the bigger context window, GPT-4o-mini (2024-07-18) or Hy3 preview?

Hy3 preview, with 262K tokens against 128K.

How many benchmarks is the GPT-4o-mini (2024-07-18) vs Hy3 preview comparison based on?

24 benchmarks have a result for both models as of October 21, 2026. GPT-4o-mini (2024-07-18)'s coding index rests on 1 independent benchmark and Hy3 preview's on 1. Where only one model has a result, the page says so instead of naming a winner.

Compare next