Coding work
Code generation, repair, and software-engineering tasks
GPT-5.6 Terra
GPT-5.6 Terra leads on the public coding lane, 64.7 to 41, with Supported evidence for both models and non-overlapping 90% intervals.
Confidence: stronger
GPT-5.2 vs GPT-5.6 Terra
Updated September 24, 2026. Rank says GPT-5.6 Terra is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Both of these models will change. Get the price, version and retirement notices for the pair, each with its source. Follow model changes
Share or export
Decision reading
GPT-5.6 Terra has the higher public score, 72.58 versus 61.37, and the 90% score intervals do not overlap. 7 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
- Shared results
- 7
- GPT-5.2 only
- 8
- GPT-5.6 Terra only
- 25
- Like-for-like categories
- 3 / 8
Supported: GPT-5.2 and GPT-5.6 TerraHow the comparison works
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Agentic work
Tool use, computer use, and multi-step task completion
GPT-5.6 Terra
GPT-5.6 Terra leads on the public agentic lane, 59.5 to 42.6, with Supported evidence for both models and non-overlapping 90% intervals.
Confidence: strongerLong documents
Prompts that approach the documented context limit
GPT-5.6 Terra
GPT-5.6 Terra has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Chat turn cost
1K fresh input + 500 output tokens
GPT-5.6 Terra
GPT-5.6 Terra has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-ratesCache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
GPT-5.6 Terra
GPT-5.6 Terra has the lower estimated token cost for this stated workload. GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Confidence: rate-fallbackRepository review cost
50K fresh input + 3K output tokens
GPT-5.2
GPT-5.2 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Which one for a specific job
Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.
The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.
41.0GPT-5.264.7GPT-5.6 Terra
Like-for-like · BenchAlign v5.7
GPT-5.6 Terra leads the like-for-like coding row.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
Same basis rules as the category table below
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
8 GPT-5.2 only
7 Shared
25 GPT-5.6 Terra only
2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
FrontierMath v2 (Tier 4)Math
Normalized gap 49.5GPT-5.2:18.800%GPT-5.6 Terra:68.300%FrontierMath v2 (Tiers 1-3)Math
Normalized gap 44.2GPT-5.2:40.700%GPT-5.6 Terra:84.900%ARC-AGI-2Reasoning
Normalized gap 31.0GPT-5.2:52.9%GPT-5.6 Terra:83.9%BrowseCompAgentic
Normalized gap 21.7GPT-5.2:65.8%GPT-5.6 Terra:87.5%SWE-bench ProCoding
Normalized gap 7.8GPT-5.2:55.6%GPT-5.6 Terra:63.4%
Bars run 0–100 on each benchmark’s normalized display scale
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign v5.7 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
Agentic
Like-for-like- GPT-5.2
- 42.6
- Supported · #42/105
- GPT-5.6 Terra
- 59.5
- Supported · #18/105
- Basis
- BenchAlign v5.7 lane · 4 vs 9 public rows
- Reading
- GPT-5.6 Terra leads
Coding
Like-for-like- GPT-5.2
- 41.0
- Supported · #54/135
- GPT-5.6 Terra
- 64.7
- Supported · #8/135
- Basis
- BenchAlign v5.7 lane · 3 vs 8 public rows
- Reading
- GPT-5.6 Terra leads
Knowledge
Like-for-like- GPT-5.2
- 57.9
- Supported · #38/158
- GPT-5.6 Terra
- 71.1
- Supported · #9/158
- Basis
- BenchAlign v5.7 lane · 1 vs 8 public rows
- Reading
- GPT-5.6 Terra leads · intervals overlap
Multimodal
Directional only- GPT-5.2
- 66.3
- #23/50
- GPT-5.6 Terra
- 77.2
- #16/50
- Basis
- Provisional lane · 2 vs 1 weighted rows
- Reading
- Directional only
Instruction following
Directional only- GPT-5.2
- 91.2
- #15/124
- GPT-5.6 Terra
- 85.7
- #35/124
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Directional only
Reasoning
Not comparable- GPT-5.2
- 60.6
- Unranked · 3 rankable rows
- GPT-5.6 Terra
- 65.4
- #12/19
- Basis
- Provisional lane · 1 vs 2 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- GPT-5.2
- Not ranked
- GPT-5.6 Terra
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- GPT-5.2
- 57.4
- Unranked · 2 rankable rows
- GPT-5.6 Terra
- 96.8
- Unranked · 3 rankable rows
- Basis
- Provisional lane · 2 vs 2 weighted rows
- Reading
- Not comparable
| Category | GPT-5.2 | GPT-5.6 Terra | Basis | Reading |
|---|---|---|---|---|
| Agentic | 42.6Supported · #42/105 | 59.5Supported · #18/105 | Like-for-likeBenchAlign v5.7 lane · 4 vs 9 public rows | GPT-5.6 Terra leads |
| Coding | 41.0Supported · #54/135 | 64.7Supported · #8/135 | Like-for-likeBenchAlign v5.7 lane · 3 vs 8 public rows | GPT-5.6 Terra leads |
| Knowledge | 57.9Supported · #38/158 | 71.1Supported · #9/158 | Like-for-likeBenchAlign v5.7 lane · 1 vs 8 public rows | GPT-5.6 Terra leads · intervals overlap |
| Multimodal | 66.3#23/50 | 77.2#16/50 | Directional onlyProvisional lane · 2 vs 1 weighted rows | Directional only |
| Instruction following | 91.2#15/124 | 85.7#35/124 | Directional onlyProvisional lane · 0 vs 0 weighted rows | Directional only |
| Reasoning | 60.6Unranked · 3 rankable rows | 65.4#12/19 | Not comparableProvisional lane · 1 vs 2 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | 57.4Unranked · 2 rankable rows | 96.8Unranked · 3 rankable rows | Not comparableProvisional lane · 2 vs 2 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.7) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Supported evidence per lane · bars run 0–100Methodology
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- GPT-5.2
- $0.00875
- Fits in one request
- GPT-5.6 Terra
- $0.008
- Fits in one request
GPT-5.6 Terra has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- GPT-5.2
- $0.1295
- Fits in one request
- GPT-5.6 Terra
- $0.136
- Fits in one request
GPT-5.2 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- GPT-5.2
- $0.525
- Fits in one request
- Cached input priced at the published list-input rate
- GPT-5.6 Terra
- $0.2
- Fits in one request
GPT-5.6 Terra has the lower modeled cost
GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Cached input falls back to the list input rate only where a cached rate is unpublished
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationGPT-5.2GPT-5.6 Terra
Context window
Maximum documented context; output-token limits may be lower.
GPT-5.2
400K
GPT-5.6 Terra
1.05M
OpenAI model catalogAPI model ID
GPT-5.2
Not sourced
GPT-5.6 Terra
gpt-5.6-terra
OpenAI model catalogCached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
GPT-5.2
Not published
GPT-5.6 Terra
$0.2 per 1M cached input tokens
OpenAI pricingDocumented inputs
GPT-5.2
Not sourced
GPT-5.6 Terra
text, image
OpenAI model catalogDocumented outputs
GPT-5.2
Not sourced
GPT-5.6 Terra
Provider availability
GPT-5.2
Not sourced
GPT-5.6 Terra
Generally Available · OpenAI Responses API
OpenAI model catalogReasoning profile
GPT-5.2
Reasoning
GPT-5.6 Terra
Reasoning
Weight access
GPT-5.2
Proprietary
GPT-5.6 Terra
Proprietary
License
GPT-5.2
Proprietary
GPT-5.6 Terra
Proprietary
Release date
GPT-5.2
2025-12-11
GPT-5.6 Terra
2026-07-09
If you already use one of these models
- Deployment change
- Both entries list OpenAI as the provider. Confirm endpoint, model ID, limits, and feature support before switching.
- Quality signal
- GPT-5.6 Terra has the higher public score, 72.58 versus 61.37, and the 90% score intervals do not overlap.
- Workload cost
- Repository review: $0.1295 vs $0.136. Cache-heavy agent loop: $0.525 vs $0.2.
- Context tradeoff
- GPT-5.6 Terra has the larger documented window (1.05M).
Run the same representative tasks against both endpoints before changing production traffic.
Questions
Which is better, GPT-5.2 or GPT-5.6 Terra?
GPT-5.6 Terra has the higher public score, 72.58 versus 61.37, and the 90% score intervals do not overlap. That is the clearest overall quality signal in the current public data.
Which is better for coding, GPT-5.2 or GPT-5.6 Terra?
GPT-5.6 Terra leads the public coding lane, 64.7 to 41, with Supported evidence for both models and non-overlapping 90% intervals.
Which is better for agentic tasks, GPT-5.2 or GPT-5.6 Terra?
GPT-5.6 Terra leads the public agentic tasks lane, 59.5 to 42.6, with Supported evidence for both models and non-overlapping 90% intervals.
Which costs less, GPT-5.2 or GPT-5.6 Terra?
For the stated presets, chat costs $0.00875 on GPT-5.2 and $0.008 on GPT-5.6 Terra; repository review costs $0.1295 and $0.136; the cache-heavy agent loop costs $0.525 and $0.2. GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Which has the larger context window, GPT-5.2 or GPT-5.6 Terra?
GPT-5.6 Terra has the larger documented context window: 1.05M, compared with 400K.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence40 rows
Agentic
BrowseComp
GPT-5.265.8%SourceGPT-5.6 Terra87.5%SourceGPT-5.6 Terra leads this result
OSWorld-Verified
GPT-5.247.3%SourceGPT-5.6 Terra—Not directly comparable
Gert Labs
GPT-5.246.54%SourceGPT-5.6 Terra—Not directly comparable
JobBench
GPT-5.234.3%SourceGPT-5.6 Terra—Not directly comparable
Terminal-Bench 3.0
GPT-5.2—GPT-5.6 Terra20.8%SourceNot directly comparable
Terminal-Bench 2.1
GPT-5.2—GPT-5.6 Terra87.4%SourceNot directly comparable
OSWorld 2.0
GPT-5.2—GPT-5.6 Terra50.2%SourceNot directly comparable
CyberGym
GPT-5.2—GPT-5.6 Terra81.8%SourceNot directly comparable
ExploitGym
GPT-5.2—GPT-5.6 Terra23.2%SourceNot directly comparable
Toolathlon
GPT-5.2—GPT-5.6 Terra53.1%SourceNot directly comparable
Terminal-Bench 2.1 (Vals)
GPT-5.2—GPT-5.6 Terra77.5%SourceNot directly comparable
ApprenticeBench
GPT-5.2—GPT-5.6 Terra16%SourceNot directly comparable
Coding
SWE-bench Verified
GPT-5.280%SourceGPT-5.6 Terra—Not directly comparable
SWE-bench Pro
GPT-5.255.6%SourceGPT-5.6 Terra63.4%SourceGPT-5.6 Terra leads this result
Vibe Code Bench
GPT-5.253.50%SourceGPT-5.6 Terra—Not directly comparable
Terminal-Bench 2.1
GPT-5.2—GPT-5.6 Terra87.4%SourceNot directly comparable
DeepSWE
GPT-5.2—GPT-5.6 Terra69.6%SourceNot directly comparable
FrontierCode 1.1 Extended
GPT-5.2—GPT-5.6 Terra55.8%SourceNot directly comparable
cursorBench32
GPT-5.2—GPT-5.6 Terra64.9%SourceNot directly comparable
VulcanBench v3
GPT-5.2—GPT-5.6 Terra87.0%SourceNot directly comparable
LiveCodeBench (Vals)
GPT-5.2—GPT-5.6 Terra85.9%SourceNot directly comparable
SWE-bench (Vals)
GPT-5.2—GPT-5.6 Terra95.4%SourceNot directly comparable
Reasoning
Multimodal
MMMU-Pro
GPT-5.279.5%SourceGPT-5.6 Terra80.7%SourceGPT-5.6 Terra leads this result
MathVision
GPT-5.283.0%SourceGPT-5.6 Terra—Not directly comparable
CharXiv
GPT-5.282.1%SourceGPT-5.6 Terra—Not directly comparable
V*
GPT-5.275.9%SourceGPT-5.6 Terra—Not directly comparable
MMMU-Pro w/ Python
GPT-5.2—GPT-5.6 Terra82%SourceNot directly comparable
Knowledge
GPQA
GPT-5.292.4%SourceGPT-5.6 Terra92.9%SourceGPT-5.6 Terra leads this result
GPQA-D
GPT-5.2—GPT-5.6 Terra92.9%SourceNot directly comparable
HLE-Verified
GPT-5.2—GPT-5.6 Terra51.1%SourceNot directly comparable
LABBench2
GPT-5.2—GPT-5.6 Terra81.2%SourceNot directly comparable
HealthBench Professional
GPT-5.2—GPT-5.6 Terra57.7%SourceNot directly comparable
HealthBench Hard
GPT-5.2—GPT-5.6 Terra32.7%SourceNot directly comparable
GPQA Diamond (Vals)
GPT-5.2—GPT-5.6 Terra90.9%SourceNot directly comparable
MMLU-Pro (Vals)
GPT-5.2—GPT-5.6 Terra86.7%SourceNot directly comparable
Math
40 public results · 7 shared
Watch GPT-5.2 vs GPT-5.6 Terra
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.
Last updated September 24, 2026