GPT-5.6 Terra

OpenAIProprietary

The balanced 5.6 tier: GPT-5.5-class intelligence at roughly half the price — where OpenAI expects everyday production traffic to land.

Index @ Medium

75.9

Context

1.05M

$/M in · out

$2.50 · $15

Median speed

150 tok/s

Released

Jul 9, 2026

02·Which effort for which task

The sweet spot, per task

Cheapest effort tier that keeps ≥96% of this model's peak quality on the benchmarks behind each task type. Below the floor number, quality is not okay — it just fails quietly.

Everyday writing & Q&A

Low

Chat, drafting, summaries, general questions.

keeps 96.5% of peak−95% cost

Code assist

Medium

Autocomplete, single functions, small fixes.

keeps 96.6% of peak−88% cost

Coding agents

High

Real repos, multi-file changes, end-to-end tickets.

keeps 98.0% of peak−65% cost

⚠ at Minimal: only 76% of peak — avoid

Math & quantitative

High

Derivations, statistics, financial models.

keeps 96.7% of peak−65% cost

⚠ at Minimal: only 63% of peak — avoid

Hard research problems

High

Novel, frontier, genuinely difficult questions.

keeps 97.5% of peak−65% cost

⚠ at Minimal: only 72% of peak — avoid

Agents & tool use

High

Long workflows with tools, policies, and state.

keeps 98.7% of peak−65% cost

03·The effort ladder

Quality and cost, tier by tier

On everyday tasks, Low keeps 96.5% of X-High quality at $0.019 vs $0.37 per task.

Intelligence Index by effort

Cost per typical task by effort

04·Benchmark detail

Scores at Minimal

Minimal Low Medium High X-High

at Minimalat X-High (peak)

Benchmark	Minimal	Peak	What it predicts
MMLU-Pro	81.8	88.0	Everyday Q&A · Drafting emails & docs · Summarizing articles
GPQA Diamond	63.4	88.0	Technical analysis · Root-cause debugging · Scientific writing
Humanity's Last Exam	24.5	34.0	Novel research questions · Frontier problem solving
AIME 2025	52.8	96.0	Financial modeling · Statistics · Algorithm design
SWE-bench Verified	60.8	78.0	Coding agents · Bug fixing in real repos · Multi-file refactors
LiveCodeBench	65.5	84.0	Writing functions from scratch · Algorithms · Code autocomplete
τ²-bench	62.9	85.0	Tool-using agents · Customer-facing automation · Multi-step workflows
IFEval	88.3	93.0	Structured output (JSON) · Templated generation · Batch pipelines

06·Drift

Is it still the same model this week?

Weekly Intelligence Index · dashed rules mark serving-change events

No serving-change events detected for GPT-5.6 Terra in the tracked window.

07·Call it

One key, this model, your effort

The gateway speaks the OpenAI Chat Completions dialect. Pick a host with gpt-5.6-terra@host syntax, set reasoning_effort, and the sweet spot above becomes one line of config. Full docs →

chat with GPT-5.6 Terra

curl https://models.fyi/api/v1/chat/completions \
  -H "Authorization: Bearer $MFYI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.6-terra",
    "reasoning_effort": "low",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
# local dev: replace https://models.fyi with http://localhost:3000