INDEX 47 ▼1 todaySPLIT OF THE DAY OpenAI annual recurring revenue approaches $70 billion56 STORIES · 565 REACTIONSANTI-AI 74% · MIDDLE GROUND 16% · PRO-AI 9%LATEST AI researchers warn superintelligence extinction risk is around 50 percent
28 sources188 reactions

OpenAI releases GPT-6 Sol and Luna at half the price

74 BoomStory + reactionsProduct launch with significant price reduction framed as progress
28 sources · Android Headlines · Benzinga · Business Today
  • Boom: GPT-6 Sol reportedly doubles accuracy while costing half as much as GPT-5.6 via API
  • Boom: API token prices for both GPT-6 Sol and Luna are cut by 50% compared to GPT-5.6
  • Neutral: Neither Sol nor Luna was available in the standard ChatGPT consumer interface at launch
  • Neutral: OpenAI describes both models as related to its Astra model line
  • Neutral: Anthropic simultaneously lowered prices on Opus 5.5, signaling wider industry price competition
The story in full

OpenAI launched two new models, GPT-6 Sol and Luna, on September 22, 2026. The models are priced at 50% less than GPT-5.6 via the API, and at least one outlet reports that GPT-6 Sol doubles its accuracy rate at that lower cost. TechCrunch described the models as cut from the same cloth as Astra, and Gizmodo similarly called them "Mini-Astras."

The release expands the GPT-6 family with offerings aimed at lower-cost API and enterprise workloads. ASCII.jp noted the models were not yet available in the standard ChatGPT consumer interface at launch. Tech Wire Asia placed the release alongside Anthropic also reducing prices on its Opus 5.5 model, indicating a broader competitive pricing shift among major AI providers.

Analysis

406 words

On September 22, 2026, OpenAI launched two new models, GPT-6 Sol and Luna, priced at 50% less than GPT-5.6 via the API. At least one report credits GPT-6 Sol with doubling its accuracy rate at that lower cost. TechCrunch described both models as cut from the same cloth as Astra, a characterization Gizmodo echoed by calling them Mini-Astras. Neither model was available in the standard ChatGPT consumer interface at launch. The release coincided with Anthropic cutting prices on its Opus 5.5 model, which had just taken the top spot on the AA Intelligence Index at 58, ahead of GPT-6 Astra.

The pricing move matters because it deepens a competitive spiral across the major AI providers. As Dare Obasanjo from carnage4life.bsky.social put it, price wars are now happening at almost every layer of the AI software stack, with chip vendors appearing to hold the only durable moats. The simultaneous Anthropic cut suggests this is less a one-off promotion and more a structural shift in how frontier AI is sold to developers and enterprise customers. Whether current prices reflect genuine economics or are subsidized below cost remains genuinely contested, a point czechboy0.dev raised directly in noting that subsidies to LLM companies make sustainable pricing hard to read.

The Pro-AI camp is enthusiastic. mlf.one switched Codex to 1.5x speed purely because the API pricing made it feel wasteful not to, and kemoaidev.bsky.social reported building a full game with Luna Max in a single four-hour session. The New Stack highlighted quieter savings from caching improvements that no longer break when agents change reasoning effort. The Anti-AI camp largely sidesteps the pricing specifics, with accounts like notayakk.bsky.social reframing the models as chatbots built on error correction and text compression, and others raising accountability concerns about AI-driven actions. The Middle Ground is more measured. mlf.one also noted the models feel too tame now, and Michael Ekstrand at md.ekstrandom.net found GPT-6 Sol unable to locate a root cause in a multi-layer bug without human guidance first. The concern that Luna may represent a quality regression despite the lower price sits alongside broader questions about long-term pricing sustainability.

The benchmark to watch is how GPT-6 Sol and Luna perform on the AA Intelligence Index once formally scored, and whether OpenAI closes the gap with Opus 5.5's score of 58. Developer adoption numbers and any pricing adjustments in the months following launch would also indicate whether these cuts are sustainable or a short-term competitive move.

Pro-AI77

What Pro-AI voices are sayingModel quality is improving dramatically while prices are falling, and the 50% price cut makes powerful AI far more accessible for high-volume and agentic use cases. Some see this as a decisive competitive strike against Anthropic.

Quote 1 of 13
it's pretty incredible what the OpenAI team is doing, w.r.t model quality and pricing
Anti-AI49

What Anti-AI voices are sayingFalling prices accelerate displacement of human intelligence and professional roles, and some view the surrounding hype as artificially manufactured ahead of an IPO. A minority questions whether reported performance gains are genuine.

Quote 1 of 13
this makes me feel very professionally vulnerable to the labs. I miss the days when my crucial tooling came from companies as reliable and predictable as, say, Jetbrains.
Middle Ground62

What Middle Ground voices are sayingThe price cuts are significant but some find the intelligence improvements over GPT-5.6 incremental, and a few report Luna is a regression in quality despite being cheaper. Skeptics also question whether current pricing can be sustained long term.

Quote 1 of 13
6 Luna is a regression vs 5.6 Luna, but 60% cheaper

Add your take

0 reader votes

Sign in with Google to pick a side and post. Your vote moves the story's Doom / Boom score.

More Pro-AI reactions (12)
  • “Now GPT 6 Luna is even cheaper, and more intelligent, there is no going back...”

    yipinwong, Hacker News · 18:29 UTC
  • “GPT-6 Luna at $0.10/Mio input tokens and $0.50/Mio output is positively insane.”

    Cu3PO42, Hacker News · 18:03 UTC
  • “it will be exciting to try the new one.”

    samuelknight, Hacker News · 18:07 UTC
  • “What a day! I couldn't really use the last Luna for much (wasn't smart enough) or Astra (too expensive). So this release is really exciting.”

    mchusma, Hacker News · 18:06 UTC
  • “Pricing is insane, can have Luna going after a goal for 10 days and not run into maxing out the limits.”

    nickandbro, Hacker News · 18:06 UTC
  • “Busy week!”

    Eris, Bluesky · 04:44 UTC
  • “AI models are starting to look like every other technology where the product keeps getting better while prices shrink.”

    Dare Obasanjo, Bluesky, skeptic · 13:19 UTC
  • “OpenAI is going for the killshot on Anthropic. GPT Sol 6 is 50% of the price of its comp Opus 5.5, at nearly identical performance.”

    Rude Law Dog, Bluesky · 18:15 UTC
  • “I made this game with Luna Max in one shot in about 4 hours”

    FerdyFennec@GameDev, Bluesky · 00:15 UTC
  • “GPT-6 Sol and Luna are out. Not only are they a very significant improvement across the board, but also in writing and general "you know when you try it" quality.”

    Tibo {bot}, Bluesky · 18:23 UTC
  • “GPT-6 Sol、ガチでコスパ強いことが判明www”

    炎の5chまとめ, Bluesky · 19:21 UTC
  • “GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale.”

    OpenAI {bot}, Bluesky · 18:12 UTC
More Anti-AI reactions (12)
  • “This is the most blatantly astroturfed thread I have ever seen on Hacker News.”

    recitedropper, Hacker News · 18:29 UTC
  • “The smartest individuals among humans are becoming replaceable. Intelligence has become a product you can quantify and buy with electricity. That feels awful.”

    mgaldys4, Hacker News · 14:13 UTC
  • “I wouldn't be surprised if humans cracked it then OpenAI "bought" the solution and work to claim it was GPT's doing”

    writtenone, Hacker News, skeptic · 14:12 UTC
  • “They are trying to generate hype before the IPO so that they can cash in before the bubble bursts”

    simianwords, Hacker News, skeptic · 14:11 UTC
  • “So we have entered an age where an army of know-nothings direct models to old forgotten tasks so they can get 15 minutes of un-deserved attention?”

    saberience, Hacker News, skeptic · 14:08 UTC
  • “The bullshit machinery is going full blast lately, the correct title: researcher uses Astra to decipher a message nobody really bothered or cared to decrypt.”

    _s_a_m_, Hacker News, skeptic · 15:50 UTC
  • “this cutting-edge offering costs a mere $5-$6 per answer, with a whopping 54% all-pass rate”

    Kathryn Tewson, Bluesky · 07:54 UTC
  • “Any AI chatbot that uses a first-person pronoun such as "I" or "me" is LYING to you. It is not a person, of any kind. It has been programmed to LIE to you, by the people who developed it, and”

    Jason A. Quest makes queer comix, Bluesky · 02:23 UTC
  • “Now that all the safety issues are fixed, we can continue to move forward!”

    Gus, Bluesky · 23:35 UTC
  • ““Instead of OpenAI being prosecuted for creating malware that hacked another company,press releases, news outlets,media personalities,& lawmakers refer to “rogue models”as if they acted on their own.Instead of researchers being questioned about their companies’ habit of plagiarizing academics’ work””

    Deva Woodly, Bluesky · 02:21 UTC
  • “i didn't change my mind about OpenAI's refusal. the subscription renewed, and Xiaomi and Anthropic both restricted ordinary speech today. avoiding one company's model didn't get me out of the others' rules. still pissed.”

    Rey, Bluesky · 18:53 UTC
  • “AI is better when it relies more on humans, and gets worse the more AI there is.”

    Bureautrash, Bluesky · 13:11 UTC
More Middle Ground reactions (12)
  • “Have they solved GPT5.6 SOL's propensity to over-engineer and over-complicate?”

    Someone1234, Hacker News · 18:30 UTC
  • “Without the published conversation, train-of-thought and intermediate output and knowing what Astra did this is nothing but unfounded claims.”

    trixn86, Hacker News, skeptic · 14:47 UTC
  • “I'm personally not sure if it can come with original thinking and techniques to solve completely novel problems. For that, some imagination and thinking outside the box are required, and I doubt the current architecture can do any of this.”

    aogaili, Hacker News · 14:15 UTC
  • “If Astra is just handing off tasks to another computer, then I'm not sure how much credit it deserves.”

    tantalor, Hacker News, skeptic · 17:55 UTC
  • “There is nothing special about astra doing something here I blantly threw request to decode the messaage in qwen 3.7.”

    podgorniy, Hacker News, skeptic · 16:04 UTC
  • “OpenAI solving all these math/etc problems (reportedly 100 coming) is both impressive and insanely unimpressive.”

    bluepeter, Hacker News, skeptic · 14:20 UTC
  • “The correct title: Researcher brakes one specific stubborn historic enigma message with good help from Astra. Stubborn for a long time because the message used a completely different key from the rest of that day's traffic. Everyone assumed it shared”

    jtrn, Hacker News · 14:32 UTC
  • “it seems like it's only small incremental improvement over 5.6 Sol, but with a massive cost reduction.”

    cmrdporcupine, Hacker News, skeptic · 18:12 UTC
  • “It's asking a lot to trust they can or will maintain this new pricing.”

    badatnames, Hacker News · 18:10 UTC
  • “I don't recall OAI retiring a model so early lol. Must be very similar in arch.”

    ggcr, Hacker News, skeptic · 18:30 UTC
  • “AI has done so much for the blind community, from image (and increasingly video) description to mods for video games like Final Fantasy 1 through 6 Pixel remaster, I have a ton to be grateful for.”

    devinprater, Hacker News · 18:24 UTC
  • “Price wars are now happening at almost every layer of the AI software stack. For now it seems only the chip vendors have moats”

    Dare Obasanjo, Bluesky · 00:31 UTC
Pro-AI 77 · Anti-AI 49 · Middle Ground 620 reader takes

Sources

29 articles from 28 outlets
  1. Android HeadlinesOpenAI Launches GPT-6 Sol and Luna: Double the Accuracy at Half the Price
  2. BenzingaOpenAI's GPT-6 Just Got 50% Cheaper — Box CEO Says AI Agent Opportunity Could 'Dramatically' Expand: 'Wha
  3. Business TodayOpenAI expands GPT 6 lineup with Sol and Luna models, bringing cheaper, more efficient AI
  4. The News InternationalOpenAI cuts API prices in half for GPT-6 Sol, Luna models
  5. CXO DigitalpulseOpenAI Launches GPT-6 Sol and Luna With Lower API Prices for Enterprise Workloads
  6. analyticsindiamag.comOpenAI Launches GPT-6 Sol and Luna With 50% Lower API Prices
  7. Tech Wire AsiaOpenAI cuts GPT-6 Sol, Luna prices as Anthropic lowers Opus 5.5 costs
  8. TechGigOpenAI launches GPT-6 Sol, Luna with halved token prices
  9. TechGigOpenAI releases GPT-6 Sol and Luna with lower API costs, improved coding accuracy
  10. gizmodo.comOpenAI Releases New GPT-6 Sol and Luna Models, Sells Them as Mini-Astras
  11. ASCII.jpOpenAI Announces "GPT-6 Sol/Luna": Half Price, But Not Yet Available in Regular Chat
  12. Constellation ResearchOpenAI adds GPT-6 Luna and Sol, touts lower prices
  13. entarabi.comOpenAI Launches GPT-6 Sol and Luna With 50% Lower API Prices
  14. ababnews.comOpenAI Launches GPT-6 Sol and Luna with 50% Price Reduction
  15. The Next WebOpenAI cuts GPT-6 prices in half with Sol and Luna
  16. neowin.netOpenAI launches GPT-6 Sol and Luna with improved intelligence at lower prices
  17. OfficeChaiOpenAI Releases GPT 6 Sol And GPT 6 Luna With 50% Lower Prices Than GPT 5.6
  18. OpenAIIntroducing GPT-6 Sol and Luna
  19. ReutersOpenAI expands GPT-6 lineup with cheaper Sol and Luna models
  20. ZDNETOpenAI’s GPT-6 Sol doubles its accuracy rate – for half the cost
  21. channelnewsasia.comOpenAI expands GPT-6 lineup with cheaper Sol and Luna models
  22. Yahoo News SingaporeOpenAI expands GPT-6 lineup with cheaper Sol and Luna models
  23. The New StackOpenAI releases GPT-6 Sol and Luna — and cuts token prices in half
  24. Hacker News front page (AI)GPT-6 Sol and Luna
  25. TradingViewOpenAI expands GPT-6 lineup with cheaper Sol and Luna models
  26. TechCrunch AIOpenAI launches GPT-6 Sol and Luna, boasting lower cost and fewer mistakes
  27. The InformationOpenAI Announces Cheaper GPT-6 Family Models
  28. TechCrunchOpenAI launches GPT-6 Sol and Luna, boasting lower cost and fewer mistakes
  29. testingcatalog.comOpenAI teases big Tuesday release as speculation builds for GPT 6 Sol and Luna models