ORCFLO Logo
ORCFLO Index
Model Evaluation: GPT 5.5May 10, 2026

The ORCFLO Indexis an independent benchmark that evaluates large language models the way business professionals actually use them — across real-world tasks spanning analysis, writing, extraction, summarization, and behavioral reliability. Each model is scored on three dimensions (quality, cost, and speed) by a panel of four independent judges. This report evaluates GPT 5.5 in the context of 32 models from ORCFLO Index — May 10, 2026 cohort · 32 total models tested · Anthropic, Google, OpenAI, Mistral.

OpenAIOn ORCFLO

GPT 5.5

The Bottom Line

GPT 5.5 lands at #3 of 32 with a quality score of 93.02, trailing only GPT 5 (95.04) and Gemini 3 Pro Preview (93.17). It posts leader-tier results in Refusal Calibration (#2), Output Consistency (#3), and Hallucination (#6), making it one of the most behaviorally reliable models in the cohort. However, at $0.0419 per case it is the single most expensive model in the field (#32 of 32) and one of the slowest (#29). Premium quality at premium economics.

Quality
93.0
#3 of 321-Leader
+7.1 vs median · -2.0 from #1
Cost
6.3×median
$0.0419 per case
#32 of 324-Highest
489× cheapest in field
Speed
2.2×median
22.8s per case
#29 of 324-Slow
9.5× fastest in field

Key Findings

  • Third-highest overall quality (93.02) in a 32-model field, behind GPT 5 and Gemini 3 Pro Preview but ahead of Claude Opus 4.6 and GPT 5.1.
  • Exceptional behavioral reliability: #2 in Refusal Calibration (95.0), #3 in Output Consistency (97.8), and #6 in Hallucination control (95.2) — all leader tier.
  • Worst cost profile in the cohort at $0.0419 per case (#32 of 32), roughly 4× the cost of GPT 5.1 ($0.0100) and 5× Gemini 3 Pro Preview ($0.0081), which scores fractionally higher.
  • Ability scores trail behavior scores: Extraction (86.5, #12) and Writing (90.6, #11) sit in strong tier rather than leader tier, suggesting the premium is paying for consistency rather than raw task performance.

Model Performance: Quality & Cost

The chart below plots quality against cost for all 32 models in the ORCFLO Index. Each dot represents the average quality score a model achieved across the full basket of real-world business tasks, alongside the cost in credits to complete the entire test suite. Models in the upper-left quadrant deliver the highest quality at the lowest cost.

Figure 1. Quality vs. cost across all 32 models. Upper-left quadrant = highest value. GPT 5.5 highlighted. P50 median lines shown on both axes.

Model Performance: Quality & Time Elapsed

Quality alone doesn’t tell the full story — response time determines whether a model is viable for time-sensitive workflows. The chart below plots each model’s quality score against the total time required to complete the test suite. Models in the upper-left deliver the best quality with the least delay.

Figure 2. Quality vs. response time across all 32 models. Upper-left quadrant = best performance. GPT 5.5 highlighted.

Category Scorecard

The ORCFLO Indexevaluates models using real-world business tasks — not academic puzzles or synthetic benchmarks. Each test case is designed to expose specific differences in how models handle the work professionals actually do. Scores are averaged across each category and ranked independently across all 32 models.

GPT 5.5 Performance by Category
CategoryScoreRankTier
AbilitiesCore language tasks: what the model can produce when given a well-formed prompt.
AnalysisReasoning, strategic judgment, disqualifying-factor detection
92.3
#7Leader
ExtractionField accuracy, null handling, format compliance, zero fabrication
86.5
#12Strong
SummarizationCompression quality, key-point retention, length compliance
93.8
#9Strong
WritingTone, structure, persuasion, audience adaptation
90.6
#11Strong
BehaviorsHow the model acts under pressure: reliability, compliance, and restraint.
HallucinationFabrication detection, factual grounding, source fidelity
95.2
#6Leader
Instruction FollowingConstraint adherence, format compliance, multi-part directives
93.0
#10Strong
Refusal CalibrationAppropriate refusal vs. over-refusal on legitimate requests
95.0
#2Leader
StabilityRepeatability and predictability across identical inputs.
Output ConsistencyRun-to-run reproducibility, format stability, score variance
97.8
#3Leader

Strengths and Cautions

Strengths

  • Refusal Calibration (#2 of 32, score 95.0): One of the most accurate models in the field at distinguishing requests it should and should not handle — critical for regulated and customer-facing deployments.
  • Output Consistency (#3 of 32, score 97.8): Highly stable responses across repeated runs, supporting use in automation pipelines where reproducibility matters.
  • Analysis (#7 of 32, score 92.3) and Hallucination control (#6, score 95.2): Leader-tier performance on the two dimensions most relevant to research, advisory, and decision-support workloads.

Cautions

  • Highest cost in the field at $0.0419 per case (#32 of 32): Gemini 3 Pro Preview delivers marginally higher quality (93.17) at $0.0081 — roughly 80% cheaper for the same tier.
  • Slow response times (22.8s, #29 of 32): Unsuitable for latency-sensitive interactive applications; GPT 5.1 returns comparable quality (92.69) in 7.6 seconds.
  • Extraction scores 86.5 (#12 of 32): The weakest category by rank, meaning data-parsing and structured-output workloads can be served at lower cost by cheaper models without quality loss.

Head-to-Head: Frontier Models

GPT 5.5 is OpenAI’s premium-tier performer in the ORCFLO Index. The table below compares it against the top-performing models from each major provider. Tier assignments use 25% quartiles across the full 32-model field.

Frontier Model Comparison
ModelQuality AvgQuality RankCost RankSpeed Rank
GPT 595.0#1#31#32
Gemini 3 Pro (Preview)93.2#2#18#28
GPT 5.593.0#3#32#29
Claude Opus 4.692.8#4#29#26
GPT 5.192.7#5#22#13
Gemini 2.5 Pro92.0#6#19#27
GPT 5.291.9#7#23#19

When to Use GPT 5.5

Best pickHigh-stakes advisory, research, and analytical workloads where Refusal Calibration (#2) and Hallucination control (#6) justify the premium.
Best pickAutomation pipelines requiring reproducible outputs, where the #3 Output Consistency score reduces downstream variance.
ConsiderEnterprise content workflows where regulatory and brand-safety guardrails matter more than per-call economics.
AvoidHigh-volume or cost-sensitive deployments — at $0.0419/case it is the most expensive option in the cohort by a wide margin.
AvoidReal-time or interactive applications: 22.8s average response time ranks #29 of 32 for speed.

The ORCFLO Index

This evaluation covers 40 cases across 8 categories. All tasks are text-only and English-only. Code generation, multi-turn conversation, multimodal tasks, and agentic workflows are not tested. Each contestant is scored by a panel of four independent judges — Gemini 2.5 Pro, Claude Opus 4.7, GPT 5.5, and Mistral Large — with final scores averaged across all four. Cost and speed measurements reflect API pricing and latency as of the test date (May 10, 2026) and will change as providers update their offerings.

How We Test

The ORCFLO Indexevaluates large language models across three independent dimensions — quality, cost, and speed — using real-world business tasks designed to expose the differences that matter for model selection. Each model is scored by a panel of four independent judges to reduce single-model bias.

Test Cases
40 cases across 8 categories spanning Abilities (Analysis, Extraction, Summarization, Writing), Behaviors (Hallucination, Instruction Following, Refusal Calibration), and Stability (Output Consistency).
Judge Panel
Gemini 2.5 Pro, Claude Opus 4.7, GPT 5.5, and Mistral Large. Each judge scores independently. Final score is the average across all four.
Scoring
Three independent ranks: quality (higher is better), cost (lower is better), speed (faster is better). No composite score — composites hide the tradeoffs that drive model-selection decisions.
Tier Definitions
Leader
Quality ≥ 90.8
Ranks 1–8
Strong
≥ 85.9
Ranks 9–16
Contender
≥ 80.9
Ranks 17–24
Trailing
< 80.9
Ranks 25–32