ZMIME
Comparison · 3 models · Updated Oct 4, 2026

GPT-4o vs Claude Haiku 3 vs Qwen2.5 72B Instruct

Claude Haiku 3 comes out ahead, 47 to 43 and 40 on our weighted score, and it is the cheaper option too.

  1. OpenAI

    GPT-4o

    Released May 13, 2024

    43/100
    • ECI129.0
    • Price$2.50 / $10.00
    • Context128K
    Full specs Remove GPT-4o
  2. Our pick

    Anthropic

    Claude Haiku 3

    Released Mar 13, 2024

    47/100
    • ECI118.4
    • Price$0.25 / $1.25
    • Context200K
    Full specs Remove Claude Haiku 3

  3. Alibaba (Qwen)

    Qwen2.5 72B Instruct

    Released Sep 19, 2024

    40/100
    • ECI129.0
    • Price$1.40 / $5.60
    • Context131K
    Full specs Remove Qwen2.5 72B Instruct
01 — Verdict

Claude Haiku 3 is our pick

Claude Haiku 3 is the better all-round choice, scoring 47/100 against GPT-4o (43) and Qwen2.5 72B Instruct (40). It leads on price and context window. GPT-4o wins on inputs & features. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.

  • CapabilityQwen2.5 72B InstructCapabilities Index (ECI): Qwen2.5 72B Instruct 129.0 · GPT-4o 129.0 · Claude Haiku 3 118.4
  • Lowest priceClaude Haiku 3Claude Haiku 3 $0.50 · Qwen2.5 72B Instruct $2.45 · GPT-4o $4.38 per 1M tokens (3:1 blend)
  • Longest contextClaude Haiku 3Claude Haiku 3 200,000 · Qwen2.5 72B Instruct 131,072 · GPT-4o 128,000 tokens
  • Widest inputsGPT-4o and Claude Haiku 3GPT-4o: Text, Images, PDFs · Claude Haiku 3: Text, Images, PDFs · Qwen2.5 72B Instruct: Text
  • Self-hostingQwen2.5 72B InstructPublishes downloadable weights
How the score is built
MeasureWeightGPT-4oClaude Haiku 3Qwen2.5 72B Instruct
CapabilityCapabilities Index (ECI)50%523852
Price25%196431
Inputs & features15%706025
Context window10%243224
Overall100%43/10047/10040/100
02 — Side by side

Every spec in one table

Highlighted cells lead their row. Dashes mean the data is not published.

GPT-4o vs Claude Haiku 3 vs Qwen2.5 72B Instruct specifications side by side
SpecificationGPT-4oOpenAIClaude Haiku 3AnthropicQwen2.5 72B InstructAlibaba (Qwen)
Capability
Capabilities Index (ECI)129.0118.4129.0 (best)
ECI rank#129 of 148#143 of 148#128 of 148 (best)
GPQA DiamondGraduate-level science questions48.9%36.3%49.2% (best)
OTIS Mock AIME 2024–2025Competition mathematics6.3%1.8%8.1% (best)
Price per million tokens
Input$2.50$0.25 (best)$1.40
Output$10.00$1.25 (best)$5.60
Cached input$1.25——
Blended (3:1)$4.38$0.50 (best)$2.45
Long-context rateSame rateSame rateSame rate
Price sourceOfficial OpenAI APIMedian of 2 providersOfficial Alibaba API
Limits
Context window128,000 tokens200,000 tokens (best)131,072 tokens
Max output16,384 tokens (best)4,096 tokens8,192 tokens
Inputs and features
TextYesYesYes
ImagesYesYesNo
PDFsYesYesNo
AudioNoNoNo
VideoNoNoNo
ReasoningNoNoNo
Tool callingYesYesYes
Structured outputYesNoNo
Availability
WeightsProprietaryProprietaryOpen
API model IDgpt-4o—qwen2-5-72b-instruct
API providers19 (best)21
ReleasedMay 13, 2024Mar 13, 2024Sep 19, 2024
Knowledge cutoffSep 2023Aug 31, 2023Apr 2024
03 — Cost

What would a month cost?

Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.

  • GPT-4o$45.00
  • Claude Haiku 3$5.00
  • Qwen2.5 72B Instruct$25.20
04 — Questions

Which should you choose?

Which is better: GPT-4o, Claude Haiku 3 or Qwen2.5 72B Instruct?

Claude Haiku 3 is the better all-round choice, scoring 47/100 against GPT-4o (43) and Qwen2.5 72B Instruct (40). It leads on price and context window. GPT-4o wins on inputs & features. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.

Which is cheaper, GPT-4o, Claude Haiku 3 or Qwen2.5 72B Instruct?

Claude Haiku 3 is cheaper at $0.25 input / $1.25 output per million tokens (median across 2 API providers). Qwen2.5 72B Instruct costs $1.40 input / $5.60 output per million tokens (official Alibaba API price); GPT-4o costs $2.50 input / $10.00 output per million tokens (official OpenAI API price). At a typical mix of three input tokens to one output token, that is $0.50 per million tokens for Claude Haiku 3 versus $2.45 for Qwen2.5 72B Instruct (4.9× as much) and $4.38 for GPT-4o (8.8× as much).

Which scores higher on benchmarks?

Qwen2.5 72B Instruct scores higher on the Capabilities Index (ECI): Qwen2.5 72B Instruct 129.0 (#128 of 148), GPT-4o 129.0 (#129 of 148) and Claude Haiku 3 118.4 (#143 of 148). The confidence ranges of the top two overlap (123.8–130.7 vs 124.2–131.5), so treat the gap as small. On individual benchmarks: GPQA Diamond — Qwen2.5 72B Instruct 49.2%, GPT-4o 48.9%, Claude Haiku 3 36.3%; OTIS Mock AIME 2024–2025 — Qwen2.5 72B Instruct 8.1%, GPT-4o 6.3%, Claude Haiku 3 1.8%.

Which is better for coding?

There are no published SWE-bench Verified results for GPT-4o, Claude Haiku 3 and Qwen2.5 72B Instruct yet, so there is no like-for-like coding score. On overall capability, Qwen2.5 72B Instruct leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.

Which has the bigger context window?

Claude Haiku 3 has the largest context window at 200,000 tokens, against 131,072 for Qwen2.5 72B Instruct and 128,000 for GPT-4o. Maximum output per response: GPT-4o up to 16,384, Claude Haiku 3 up to 4,096, Qwen2.5 72B Instruct up to 8,192 tokens.

Which can read images, PDFs, audio or video?

GPT-4o accepts text, images and PDFs; Claude Haiku 3 accepts text, images and PDFs; Qwen2.5 72B Instruct accepts text. GPT-4o handles the widest range of inputs.

Are any of these open source?

Qwen2.5 72B Instruct publishes its weights and can be self-hosted; GPT-4o and Claude Haiku 3 is proprietary.

Which is newer?

Qwen2.5 72B Instruct is the newest, released Sep 19, 2024. GPT-4o came out May 13, 2024; Claude Haiku 3 came out Mar 13, 2024. Knowledge cutoff: GPT-4o Sep 2023, Claude Haiku 3 Aug 31, 2023, Qwen2.5 72B Instruct Apr 2024.

How do you decide the winner?

Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.

05 — Keep comparing

Related comparisons