Side-by-side comparison of pricing, 12 benchmarks, and generation speed.
| Metric | Claude 4.5 Haiku (Reasoning) | GPT-5.4 mini (xhigh) |
|---|---|---|
| Input ($/M tokens) | $1 | $0.75 |
| Output ($/M tokens) | $5 | $4.5 |
Verdict. GPT-5.4 mini (xhigh) takes the aggregate benchmark matchup 2–0 across 2 categories. Real workloads usually care about a handful of specific tasks — see the per-benchmark table above.
Pricing. Both models sit in the mid-tier bracket for output-token pricing. At 1.1× the per-million-token cost, GPT-5.4 mini (xhigh) is meaningfully cheaper if your traffic is output-heavy (long completions, document generation, agent loops). GPT-5.4 mini (xhigh) makes more sense when output volume is low and absolute reasoning quality justifies the premium.
Strengths. Claude 4.5 Haiku (Reasoning) is strongest on Coding Index (43.9), Intelligence Index (29.9). GPT-5.4 mini (xhigh) leads on Coding Index (56.1), Intelligence Index (40.9).
Speed. On throughput, GPT-5.4 mini (xhigh) generates tokens at 163 tok/s versus 89 tok/s — about 45% faster. On time-to-first-token, GPT-5.4 mini (xhigh) responds in 11890ms vs 16690ms, which matters most for chat-style UIs.
Provider. Anthropic and OpenAI sell to overlapping but distinct developer audiences: Anthropic tends to ship frontier reasoning models with premium positioning, while OpenAI often prices more aggressively. Your existing vendor relationships, billing, and SLA preferences may matter as much as the raw numbers above.
Workload cost. Workload scenarios (per million requests at 30M input + 15M output tokens): Claude 4.5 Haiku (Reasoning) costs $105.00 ($1260/year); GPT-5.4 mini (xhigh) costs $90.00 ($1080/year). At a smaller 5M-input/2M-output scale (single-developer tool or prototype): Claude 4.5 Haiku (Reasoning) ≈ $15.00/run, GPT-5.4 mini (xhigh) ≈ $12.75/run. At agent/realtime scale (200M input / 100M output per million requests): Claude 4.5 Haiku (Reasoning) ≈ $700/run, GPT-5.4 mini (xhigh) ≈ $600/run. GPT-5.4 mini (xhigh) becomes more attractive at higher volume — the absolute per-token pricing difference compounds when you ship at scale.
Recommendation. Both models have legitimate use cases — the right answer depends on whether you are optimizing for benchmark ceiling, latency, or unit cost. Start with the cheaper / faster model, evaluate against your specific task, and only switch if the upgrade shows a meaningful lift.
Head-to-head deltas
Data from Artificial Analysis API — 12 benchmarks
GPT-5.4 mini (xhigh) is cheaper overall. Its blended price (3:1 input/output ratio) is $1.69/M tokens vs $2.00/M for Claude 4.5 Haiku (Reasoning).
GPT-5.4 mini (xhigh) wins 2 out of 12 benchmarks compared to 0 for Claude 4.5 Haiku (Reasoning). See the detailed benchmark chart above for per-category results.
GPT-5.4 mini (xhigh) generates tokens faster at 163 tok/s vs 89 tok/s. However, GPT-5.4 mini (xhigh) has lower time-to-first-token (11.89s vs 16.69s).
Choose based on your priorities: GPT-5.4 mini (xhigh) for lower cost, GPT-5.4 mini (xhigh) for stronger benchmark performance, and GPT-5.4 mini (xhigh) for faster generation. For latency-sensitive apps, check the TTFT comparison above.