Side-by-side benchmark comparison across agentic, coding, multimodal, knowledge, reasoning, and math workflows.
GPT-5.4 has the cleaner overall profile here, landing at 90 versus 87. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
GPT-5.4's sharpest advantage is in multimodal & grounded, where it averages 95.5 against 88.3. The single biggest benchmark swing on the page is MMMU-Pro, 95 to 86.
GPT-5.4 is also the more expensive model on tokens at $2.50 input / $15.00 output per 1M tokens, versus $2.00 input / $8.00 output per 1M tokens for GPT-5.3-Codex-Spark. GPT-5.4 gives you the larger context window at 1.05M, compared with 256K for GPT-5.3-Codex-Spark.
Pick GPT-5.4 if you want the stronger benchmark profile. GPT-5.3-Codex-Spark only becomes the better choice if you want the cheaper token bill.
GPT-5.4
87.8
GPT-5.3-Codex-Spark
85.6
GPT-5.4
84.5
GPT-5.3-Codex-Spark
82.3
GPT-5.4
95.5
GPT-5.3-Codex-Spark
88.3
GPT-5.4
95.8
GPT-5.3-Codex-Spark
92.7
GPT-5.4
82.6
GPT-5.3-Codex-Spark
78.3
GPT-5.4
96
GPT-5.3-Codex-Spark
92
GPT-5.4
94.7
GPT-5.3-Codex-Spark
90.8
GPT-5.4
98.2
GPT-5.3-Codex-Spark
96.7
GPT-5.4 is ahead overall, 90 to 87. The biggest single separator in this matchup is MMMU-Pro, where the scores are 95 and 86.
GPT-5.4 has the edge for knowledge tasks in this comparison, averaging 82.6 versus 78.3. Inside this category, HLE is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for coding in this comparison, averaging 84.5 versus 82.3. Inside this category, HumanEval is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for math in this comparison, averaging 98.2 versus 96.7. Inside this category, AIME 2025 is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for reasoning in this comparison, averaging 95.8 versus 92.7. Inside this category, MRCRv2 is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for agentic tasks in this comparison, averaging 87.8 versus 85.6. Inside this category, BrowseComp is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for multimodal and grounded tasks in this comparison, averaging 95.5 versus 88.3. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for instruction following in this comparison, averaging 96 versus 92. Inside this category, IFEval is the benchmark that creates the most daylight between them.
GPT-5.4 has the edge for multilingual tasks in this comparison, averaging 94.7 versus 90.8. Inside this category, MMLU-ProX is the benchmark that creates the most daylight between them.
Get notified when new models drop, benchmark scores change, or the leaderboard shifts. One email per week.
Free. No spam. Unsubscribe anytime. We only store derived location metadata for consent routing.