Tech Meridian← ALL MODELS
PROMY MERIDIAN

OPENAI · MODEL RELEASE TRACKER

GPT-6 Luna

GPT-6 Luna is an OpenAI GPT-6 variant designed as a low-cost software agent for healthcare administrative workflows. It operates by interacting with existing user interfaces (mouse clicks and keystrokes), was integrated into the Agents API in late September, is available via the API and in ChatGPT for Healthcare/Clinicians, and scored 91.8% on the HealthAdminBench benchmark. OpenAI reported a cost of about 8.5 cents per task and US deployment including use at CVS Health.

CURRENT SNAPSHOT4/5 DIMENSIONS WITH DATA

The dimensions that change the decision.

PRICING
API pricing per administrative taskAbout $0.085 (8.5 cents) per task; reported as 98% less than GPT-6 Astra.
Price change versus previous generationReportedly priced at about half the token price of the GPT-5.6 Sol and Luna models (≈50% reduction reported).
CONTEXT WINDOW

Not established from the available sources.

MODALITIES
Interface interaction modalityOperates clinic programs and insurance portals by interacting with existing user interfaces using mouse clicks and keystrokes.
Input modalitiesAccepts text and image input.
BENCHMARKS
Reported benchmark/performance change versus predecessorOne report states Luna 'improves on its predecessor by 5.4 percentage points while costing 58 percent less' on the evaluations referenced by that coverage; the excerpt does not name the specific benchmark.
Internal benchmark vs GPT‑5.6In internal tests using the streaming method, GPT‑6 scored higher than its predecessor GPT‑5.6 on difficult web searches.
MultiLangBench score (legacy/low-resource languages)On OpenAI's MultiLangBench evaluation (maximum reasoning effort), GPT-6 Luna scored 40.3 percent.
Reported token-generation throughput on Nvidia GPUsIn a third-party comparison published by Inception, GPT-6 Luna is reported to generate about 130 tokens per second on Nvidia GPUs (Inception's internal throughput comparison).
Performance vs. Inception's Mercury 3Inception's internal benchmarks report that their Mercury 3 model outperformed GPT-6 Luna in eight of 12 tests and led GPT-6 Luna by 15 points on IFBench in the company's comparison.
Coding Agent Index v2.0 scoreScored 31 on the Coding Agent Index v2.0 (reported value: 31).
Benchmark evaluation (StateBench)GPT-6 Luna was evaluated on StateBench, a benchmark for long-context state tracking.
HealthAdminBench benchmark scoreScored 91.8% on the HealthAdminBench benchmark; ran at maximum reasoning effort during the benchmark.
AVAILABILITY
Availability channelsFeatures are available through the Agents API and in ChatGPT for Healthcare and ChatGPT for Clinicians.
Reported enterprise adoptionUS companies, including CVS Health, are reported to use the agents to process referrals and specialist consultations across separate systems.

RELEASE TIMELINE

Published, source-backed release events only.

1 SOURCES · IMPORTANCE 8.0

Anthropic launches Claude Haiku 5.5 with big price cuts and benchmark gains

Anthropic released Claude Haiku 5.5, a faster, lower‑cost small model aimed at high‑volume tasks; Anthropic says Haiku 5.5 costs ~75% less on average than Haiku 4.5 and up to 90% less for prompts ≤100,000 tokens. The company also halved Sonnet 5.5 cache‑read costs, added monthly API credits for subscribers, updated SDKs with beta computer/browser use, and says Haiku 5.5 outperforms its predecessor (and compares favorably to OpenAI’s GPT‑6 Luna) on several benchmarks and agentic coding tests.

→
3 SOURCES · IMPORTANCE 9.0

OpenAI rolls out GPT-6 with Intelligent UI in ChatGPT

GPT-6 is rolling out globally inside ChatGPT and introduces an "Intelligent UI" that delivers faster responses along with visuals and interactive experiences you can explore and use directly.

→
1 SOURCES · IMPORTANCE 7.0

OpenAI reopens $200 Pro plan but halves API credits per dollar

OpenAI is reopening its $200-per-month Pro subscription to new sign-ups while cutting the amount of API credits granted per dollar by half; the previous 5-hour usage cap for subscribers has been removed. Employee Thibault Sottiaux cited recent releases—GPT-6 Sol and GPT-6 Luna, reportedly released this week at half the API price of predecessors—and framed the changes as a shift toward pay‑per‑use pricing.

→

VERIFIABLE FACTS

Every value stays attached to a source and date.

CAPABILITIES · Intended use-cases / capabilitiesDEVELOPER CLAIM

Positioned for high-volume, well-defined, repeatable tasks — examples given include summarizing documents, extracting information, and answering short questions. Coverage describes Luna as making focused, repeatable tasks practical at high volume where small differences in latency and cost matter.

LIMITATIONS · Performance note / limitation reported by coverageDEVELOPER CLAIM

Some coverage characterizes Sol and Luna as largely matching their predecessors' performance while delivering significant price reductions (phrases used include 'barely move the needle on performance'), indicating that the primary change in these releases was cost efficiency rather than large capability gains.

LIMITATIONS · Relative rule violations in diet-planning evaluationINDEPENDENTLY SUPPORTED

In the paper's dietitian-approved weekly planning evaluation, a locally hosted 14B planner incurred fewer rule violations than its GPT-6 Luna teacher, indicating GPT-6 Luna had higher rule violations in that evaluation.

WHAT CHANGED

Stored passport versions, without reconstructed history.

Passport updated34 facts
RELEASE · Agents API integration+ The computer-use agents (including Luna) were integrated into the Agents API in late September, enabling them to operate existing UIs the way a human would.
AVAILABILITY · Availability channels+ Features are available through the Agents API and in ChatGPT for Healthcare and ChatGPT for Clinicians.
AVAILABILITY · Availability in ChatGPT clients− At launch, Luna is reported to be available in the ChatGPT desktop app and to Free and Go users; broader ChatGPT availability was described as rolling out gradually.
CAPABILITIES · Design emphasis+ Built for low API pricing (described by OpenAI as a lower-cost agent option compared with Astra).
AVAILABILITY · Reported enterprise adoption+ US companies, including CVS Health, are reported to use the agents to process referrals and specialist consultations across separate systems.
AVAILABILITY · Availability on Google Cloud (private preview)− GPT-6 Luna is offered on Google Cloud as part of a set of OpenAI models in a private preview for select enterprise customers; sign-ups are open and the models will also appear in Gemini for Business and Google Antigravity. Charges run through a single Google Cloud bill and count toward existing contracts.
MODALITIES · Interface interaction modality+ Operates clinic programs and insurance portals by interacting with existing user interfaces using mouse clicks and keystrokes.
CAPABILITIES · Medical administrative task handling (tests)+ In tests, reliably handled medical administrative tasks such as transferring data between electronic health records and external web portals.
BENCHMARKS · HealthAdminBench benchmark score+ Scored 91.8% on the HealthAdminBench benchmark; ran at maximum reasoning effort during the benchmark.
RELEASE · Rollout with Intelligent UI− GPT‑6 is rolling out globally in ChatGPT with a new Intelligent UI that provides visuals and interactive experiences.
MODALITIES · Modality− Text-based conversational LLM participating in extended multi-turn dialogues (used in an LLM-LLM debate framework).
PRICING · Positioning (price sensitivity)− Reported as a very price-sensitive (lower-cost) option — recommended for users who are very price sensitive.
PRICING · API pricing per administrative task+ About $0.085 (8.5 cents) per task; reported as 98% less than GPT-6 Astra.
Passport updated31 facts
BENCHMARKS · Benchmark evaluation (StateBench)+ GPT-6 Luna was evaluated on StateBench, a benchmark for long-context state tracking.
LIMITATIONS · Limitation: degrades on harder state-tracking with longer inputs+ On StateBench, GPT-6 Luna "still breaks down on harder forms of state tracking as inputs grow longer."
Passport updated29 facts
LIMITATIONS · Relative rule violations in diet-planning evaluation+ In the paper's dietitian-approved weekly planning evaluation, a locally hosted 14B planner incurred fewer rule violations than its GPT-6 Luna teacher, indicating GPT-6 Luna had higher rule violations in that evaluation.
Passport updated28 facts
BENCHMARKS · Coding Agent Index v2.0 scoreReportedly scored 31 on the Coding Agent Index v2.0 (17 points below GPT-6.1 Terra's 48).→Scored 31 on the Coding Agent Index v2.0 (reported value: 31).
PRICING · Positioning (price sensitivity)+ Reported as a very price-sensitive (lower-cost) option — recommended for users who are very price sensitive.
PRICING · Relative cost compared with other GPT-6 variants− Reports describe GPT-6 Luna as a lower-cost alternative compared with higher‑performing GPT-6 variants such as GPT-6 Astra.
Passport updated28 facts
BENCHMARKS · Coding Agent Index v2.0 score+ Reportedly scored 31 on the Coding Agent Index v2.0 (17 points below GPT-6.1 Terra's 48).
Passport updated27 facts
MODALITIES · Interactive visuals and UI elements− GPT‑6 can present answers as interactive interfaces that combine text with diagrams, charts, forms, tappable buttons, graphics, and small built‑in tools (mini apps).
MODALITIES · Modality+ Text-based conversational LLM participating in extended multi-turn dialogues (used in an LLM-LLM debate framework).
CAPABILITIES · Observed opinion-dynamics behavior+ In experiments involving GPT-6 Luna, same-model interactions tended to narrow the opinion gap on average, while cross-model interactions tended to widen it; conversational influence depended on the participating models.
LIMITATIONS · Effect of blanket persuasion instruction+ A blanket instruction to persuade did not reliably increase a model's influence in the LLM-LLM debate experiments involving GPT-6 Luna.
Passport updated25 facts
AVAILABILITY · Availability on Amazon Bedrock (global cross-Region inference)− Available through Amazon Bedrock global cross-Region inference via the bedrock-runtime endpoint.
SAFETY · Used to classify harmful commands− Vercel previously used GPT-6 Luna to classify harmful commands; it was later replaced by Jev for that role.
SAFETY · Customer data usage policy when hosted on Google Cloud+ Google states that customer data submitted to OpenAI models (including GPT-6 Luna) via Google Cloud will not be used for training.
AVAILABILITY · Availability on Google Cloud (private preview)+ GPT-6 Luna is offered on Google Cloud as part of a set of OpenAI models in a private preview for select enterprise customers; sign-ups are open and the models will also appear in Gemini for Business and Google Antigravity. Charges run through a single Google Cloud bill and count toward existing contracts.
PRICING · Free/budget variant− Paying customers receive GPT‑6 Sol, while free users receive GPT‑6 Luna; media outlets refer to GPT‑6 Luna as OpenAI's budget/free variant.
BENCHMARKS · Performance vs. Inception's Mercury 3+ Inception's internal benchmarks report that their Mercury 3 model outperformed GPT-6 Luna in eight of 12 tests and led GPT-6 Luna by 15 points on IFBench in the company's comparison.
BENCHMARKS · MultiLangBench score (legacy/low-resource languages)+ On OpenAI's MultiLangBench evaluation (maximum reasoning effort), GPT-6 Luna scored 40.3 percent.
CAPABILITIES · Positioning — fast, cost-efficient option+ OpenAI positions GPT-6 Luna as a fast, cost-efficient option for everyday use.
PRICING · Relative cost compared with other GPT-6 variants+ Reports describe GPT-6 Luna as a lower-cost alternative compared with higher‑performing GPT-6 variants such as GPT-6 Astra.
BENCHMARKS · Reported token-generation throughput on Nvidia GPUs+ In a third-party comparison published by Inception, GPT-6 Luna is reported to generate about 130 tokens per second on Nvidia GPUs (Inception's internal throughput comparison).
Passport updated21 facts
SAFETY · Used to classify harmful commands+ Vercel previously used GPT-6 Luna to classify harmful commands; it was later replaced by Jev for that role.