LLM Compare
Head to head

Gemini 4 Carbon vs Hy4 flash

How Gemini 4 Carbon and Hy4 flash stack up across benchmarks, pricing, speed, and the workloads that matter.

Gemini 4 Carbon

Google · Proprietary

Blended price$2.89per 1M tokens
Context1.05Mtokens
Speed11.3tok/s
Benchmarks16results
VS

Hy4 flash

Tencent · Proprietary

Blended price$0.09per 1M tokens
Context960Ktokens
Speedn/atok/s
Benchmarks9results

Gemini 4 Carbon costs $2.89 per million blended tokens, about 30.93x the price of Hy4 flash. On coding, Gemini 4 Carbon leads by about 37 points in our normalized benchmark average.

The table below breaks down every workload axis where both models have published results. Each score is a normalized 0 to 1 average across the benchmarks tagged with that category. A gap of a few points is noise; a gap of ten or more is a real difference in capability. Where one model has not reported a result, the cell shows n/a and the missing benchmark does not drag its average down.

Benchmark scores by workload

WorkloadGemini 4 CarbonHy4 flashEdge
Reasoning 77% 60% +16.9 pts
Coding 70% 33% +37.1 pts
Agents & tool use 70% 41% +29 pts
Math 80% 55% +24.8 pts
Vision & multimodal 81% 62% +19.1 pts
Long context 96% n/a n/a
Knowledge & factuality 72% 72% +0.2 pts

The headline benchmarks below are the most widely cited individual tests. GPQA Diamond measures graduate-level reasoning in physics, chemistry, and biology. SWE-Bench Verified tests whether a model can fix real GitHub issues. MMMU-Pro covers college-level multimodal understanding across six disciplines. AIME and FrontierMath push competitive and research math. BrowseComp measures web research ability.

Headline benchmarks

BenchmarkGemini 4 CarbonHy4 flash
MMMU-Pro n/a 76%

Benchmarks tell you what a model can do in a controlled setting. They do not tell you whether it will work on your specific task, with your specific data, at your specific scale. The recommendations below map each common workload to whichever of these two models scores higher on the relevant axis. Use them as a starting point, not a final answer.

Which should you choose?

Hard reasoning

Multi-step analysis, research, and problems that need sustained thought.

Gemini 4 Carbon

Coding & software

Writing, reviewing, and debugging code across a real codebase.

Gemini 4 Carbon

Agents & tool use

Long-running agents that call tools, browse, and act on their own.

Gemini 4 Carbon

Math & science

Formal math, competitive problems, and quantitative science.

Gemini 4 Carbon

Vision & documents

Reading images, screenshots, charts, and dense documents.

Gemini 4 Carbon

Long documents

Whole codebases, books, and transcripts that fill the context window.

Gemini 4 Carbon

High-volume & cost-sensitive

Cheap, repetitive calls where the blended token price dominates.

Hy4 flash

Scores are normalized from public benchmarks published by llm-stats.com and averaged per workload. Pricing is the blended input/output cost per million tokens at an 8:1 mix. Refreshes daily. How this works.