GPT-6.1 Sol Instantly Expands to 500k Context, Completely Free

marsbitPublished on 2026-10-13Last updated on 2026-10-13

Abstract

GPT-6.1 Sol has expanded its default context window to 500k tokens, an 80% increase from the previous 272k limit. This update aims to reduce the frequent and slow context compression that developers faced during long coding tasks, allowing the model to retain more details. When using subscription credits, OpenAI has also removed the additional long-context fee for this model, though standard credit consumption for tasks still applies, and API pricing remains separate. However, some developers question whether this is a substantial upgrade or merely a configuration change. Concerns remain about whether the larger context will lead to faster credit depletion due to increased cache reads, even without extra fees. Additionally, while GPT-6.1 Sol supports up to 1M tokens manually, its default 500k setting is positioned against competitors like Claude Code, whose Opus 5.5 and Sonnet 5.5 models already offer a default 1M context window with standard token pricing and their own compression strategies. OpenAI indicates plans for further context window expansions in the future.

GPT-6.1 Sol's default context window has surged directly to 500k!

With the OpenAI Day 7 update, the default window in Codex and ChatGPT Work expanded from 272k to 500k, a jump of about 80% in one go.

Immediately after, Tibo added another point: Using subscription credits to run GPT-6.1 Sol will not incur long-context surcharges.

Tasks will still consume subscription credits as usual, but there will be no extra charge specifically for the longer context.

API calls are still billed according to the API pricing table.

However, developer Mhmd Hammoud countered in the comments: "Does tweaking a configuration now count as a real upgrade?"

GPT-6.1 Sol Roars to 500k, Wish Granted

Even before the default 500k was rolled out, developer Lasse was already fed up with waiting for compression.

He once complained, "Codex compressing the context could make me wait up to two minutes at worst."

Two days before this update, he simply set his window to 500k himself, posted a screenshot of the status, and called on OpenAI to fix the slow compression issue quickly.

When asking Codex to review files, modify code, or run tests, the code read and errors encountered all occupy the context.

When the context is nearly full, it must first compress and organize the earlier conversation and progress to free up space before it can continue working.

After waiting for compression, you still have to check if it missed the initial requirements.

For example, you might have said, "This interface needs to be backward compatible." If it modifies a dozen files, the new feature works, but the old interface breaks, you'd end up having to fix it yourself.

Tibo stated that this expansion reduces the need for compression and allows the model to retain more details during long tasks.

Fewer waits for compression, fewer repeats of the same requirement – that's how the 500k truly helps.

Surcharge Waived, Credits Still Deducted

The biggest fear was that less compression might lead to credits depleting even faster. OpenAI had indeed stepped into this pitfall before.

In July of this year, OpenAI increased the context in their products from 272k to 372k for use with GPT-5.6 Sol.

The result was subscription usage being deducted more than expected, so the team rolled back to 272k. The capacity just added was promptly taken away.

Developer Theo speculated at the time that it might be due to double charging for context exceeding 272k.

Tibo corrected this, saying that subscriptions back then also didn't have a long-context surcharge. The extra usage came from repeatedly reading more cached content.

You only typed "continue," but the Agent, needing to continue modifying code, still has to bring along the previously read files, test results, and requirements.

When subsequently calling tools and continuing reasoning, this content might be read repeatedly.

Caching allows reuse of already processed content at a lower read cost, but each read still counts towards usage. More content read and more frequent reads accumulate consumption.

Waiving the surcharge doesn't necessarily mean the same task will definitely use fewer credits.

Developers want to wait less for compression, but not run out of credits before the project is finished.

Claude Already Defaults to a Million

However, whether 500k is sufficient must be compared with Claude.

In Claude Code, Opus 5.5 and Sonnet 5.5 already have a default context of 1M, double the new default value of Codex.

In reality, GPT-6.1 Sol's upper limit can also reach 1M, and users can manually increase it.

However, OpenAI defaulted to 500k this time because, after testing, they believe this setting is more suitable for most users.

Regarding costs, when calling the API for these two Claude models, the standard token price applies within 1M, and the unit price doesn't increase with longer context.

What OpenAI waived this time is the long-context surcharge within subscription credits; API pricing is separate. Both sides are referring to different fee scopes.

Even with a 1M window, it can still fill up. In Claude Code, using these two models, the default auto-compression point is around 967k.

When nearing the limit, it first removes earlier tool outputs from the context, such as records from reading files or running commands.

If more space is needed, it then condenses the conversation into a summary, retaining the current requirements and key code, allowing the task to continue.

Sub-tasks like file lookups can be delegated to a sub-Agent. It has its own context, and after the lookup, it returns a summary of the results.

This way, the main Agent can proceed with the results without needing to stuff the entire lookup process into the main conversation.

OpenAI states that the default window will continue to increase in the future.

The window will grow larger, but developers' patience cannot expand alongside it. After handing code to AI, there should be fewer waits and fewer repeats of "I mentioned this requirement earlier."

References:

https://x.com/thsottiaux/status/2109430343346581527?s=20

This article is from the WeChat public account "新智元" (New Zhiyuan), author: Moses & Taozi

Trending Cryptos

Related Questions

QWhat is the new default context window size for GPT-6.1 Sol announced by OpenAI, and by what percentage was it increased?

AThe new default context window size for GPT-6.1 Sol is 500k, representing an increase of approximately 80% from the previous 272k.

QAccording to the article, what key change did OpenAI make regarding fees when using subscription credits to run GPT-6.1 Sol?

AOpenAI announced that using subscription credits to run GPT-6.1 Sol will no longer incur an additional long-context fee. Tasks consume the standard subscription credits but without an extra charge for the longer context.

QWhat was the primary user complaint that the expansion of the context window to 500k aimed to address?

AThe expansion aimed to address user frustration with slow context compression, where compressing the context could take up to two minutes, interrupting workflows and requiring users to re-state instructions.

QHow does the default context window of Claude's Opus 5.5 and Sonnet 5.5 models compare to GPT-6.1 Sol's new default?

AClaude's Opus 5.5 and Sonnet 5.5 models in Claude Code have a default context window of 1M (one million tokens), which is twice the size of GPT-6.1 Sol's new default of 500k.

QWhat was the reason given by OpenAI for setting the default context window to 500k instead of the maximum 1M that GPT-6.1 Sol supports?

AOpenAI set the default to 500k because, after testing, they determined this setting was more suitable for the majority of users, balancing performance and cost-effectiveness.

Related Reads

After Jensen Huang's Departure, 14 Investment Banks Rush to Interpret: No One Can Fill His Leather Jacket, Bottom Consensus Around $150

Nvidia announced that co-founder and CEO Jensen Huang has resigned immediately, along with his board seat. CFO Colette Kress becomes interim CEO, with Harvey Jones as chairman and a permanent CEO search underway. Shares fell about 10% premarket to around $156, below the 52-week low, wiping out roughly $420 billion in market value; the stock is down about a third over the past month. Of 14 banks issuing research, three downgraded the stock (Bernstein, KeyBanc, William Blair) and six cut targets (Bernstein, JPMorgan, RBC, Cantor, Goldman, Citi), though most kept Buy ratings. The average target is about $260, implying roughly 50% upside. The consensus: earnings estimates are unchanged, but the multiple is compressed—Goldman cut its target from $300 to $220, keeping $10 EPS but lowering the multiple from 30x to 22x, likening the situation to Apple after Steve Jobs (though noting Apple had a succession plan and no investigations). Bulls argue orders are sold out through 2028, with the roadmap set beyond that, so fundamentals are safe near-term; RBC sees demand exceeding supply by ~50%. Bernstein's Stacy Rasgon downgraded to Market Perform, cutting his target from $400 to $200, warning that "if you see two ants in the kitchen," there are likely more issues. Key concerns: organizational structure, with ~60 executives reporting directly to Huang and no clear successor—"no one can fill Jensen's leather jacket." Ian Buck is the most-mentioned internal candidate. Another worry is Nvidia's role as the AI ecosystem's financier, injecting $50-60 billion in quarterly free cash flow into OpenAI, Anthropic, CoreWeave and suppliers; UBS warns of a possible "pause" that could pull $50 billion from the system. Regulatory scrutiny continues (DOJ, House, Senate). Most banks see a bottom around $150, with RBC's bear case at $100. Competitors AMD and Broadcom are seen as beneficiaries.

marsbit12m ago

After Jensen Huang's Departure, 14 Investment Banks Rush to Interpret: No One Can Fill His Leather Jacket, Bottom Consensus Around $150

marsbit12m ago

Nobel Laureate Hassabis Delivers Another Masterpiece: AlphaGenome 2 Deciphers 98% of Human DNA 'Dark Matter'

DeepMind has unveiled AlphaGenome 2, a next-generation DNA foundation model led by Nobel laureate Demis Hassabis. While AlphaFold decoded only the 2% of the human genome that encodes proteins, AlphaGenome 2 targets the remaining 98% of non-coding "dark matter," which harbors gene-regulatory switches and over 90% of disease-causing mutations. Released alongside a 78-page technical report, the model delivers four major breakthroughs. First, it achieves single-cell resolution, trained on over 107 million single cells across 68 human tissues, doubling the rate at which it identifies pathogenic mutations—raising detection from 16% to 30% and directional accuracy from 23% to 43% on the TenK10K immune-cell benchmark. Second, it reproduces years of Nature-published research purely through computation: with zero wet-lab experiments, it correctly identified a distant regulatory mutation that over-activates ETS2 across 290,000 bases to cause inflammatory bowel disease, and precisely located the regulatory site and GATAA core sequence targeted by approved sickle-cell CRISPR therapy. Third, a new cell encoder lets the model interpret previously unseen diseases from just dozens of patient cells without retraining—identifying 41% of pathogenic mutations in unseen Parkinson's brain samples versus 17% for a specialized model, and uncovering a hidden severe-COVID mutation active only in immune-activated T cells. Finally, DeepMind released free databases covering 640 real human cell types and impact scores for nearly 7 million genetic variants across 510 cell states, with APIs and Colab tutorials open for non-commercial research. The model will soon be packaged as Skills and integrated into Google Antigravity for direct use by Gemini 4, promising to dramatically shorten the cycle of disease investigation and drug discovery.

marsbit1h ago

Nobel Laureate Hassabis Delivers Another Masterpiece: AlphaGenome 2 Deciphers 98% of Human DNA 'Dark Matter'

marsbit1h ago

Midterm Elections Settled, What Lies Ahead for the CLARITY Act?

The CLARITY Act (Digital Asset Market Clarity Act), which had already struggled before the election, now faces dim prospects after Democrats swept both chambers in the midterms. On September 15, a Senate cloture motion on the House-passed bill failed 49-50, eleven votes short of the 60-vote threshold. All 49 yes votes came from Republicans; the opposition included all Democrats present, two independents, and four Republicans (Collins, Hawley, Moran, Tillis). Even if absentee Democrat Coons had voted yes and the four Republicans had flipped, the tally would have reached only 54—still short. The bill's core compromise would make the CFTC the primary regulator for digital commodities and trading platforms while preserving clear SEC authority; jurisdictional division and added ethics provisions are the main disputes. Democrats will control both chambers when the new Congress opens January 3, 2027, leaving little incentive to cooperate during the lame-duck session. Only two theoretical paths remain before the 119th Congress ends: renegotiate the text and hold another cloture vote via Tillis's reconsideration motion, or attach crypto market structure provisions to must-pass December legislation such as government funding (expiring December 11) or the NDAA. Both are unlikely, given the tight legislative calendar and Democratic leverage. Stablecoin oversight continues separately under the GENIUS Act and OCC rulemaking. If the bill dies, it must be reintroduced in the 120th Congress, where Democrats—likely led by Elizabeth Warren and Maxine Waters—are expected to draft a very different framework expanding SEC authority and narrowing the CFTC's role. Key things to watch before January 3: any new cloture vote or revised text, any attempt to attach provisions to funding or defense bills, outstanding vote counts in Alaska, Mississippi, and House districts, and the incoming committee chair selections. Absent such developments, the CLARITY Act is effectively finished as a 2026 legislative project.

marsbit1h ago

Midterm Elections Settled, What Lies Ahead for the CLARITY Act?

marsbit1h ago

4.8 Billion Dollars: Tether's Hand Has Reached Into 13 U.S. Public Companies

Tether, the world's largest stablecoin issuer, holds roughly $4.8 billion in public-market exposure across 13 U.S.-listed companies, according to a Protos review of regulatory filings. Its holdings span a controversial video platform, gold royalty companies, a satellite firm, bitcoin miners, and a bitcoin treasury project. Key positions include a 50.3% stake (including warrants) in Rumble (RUM), though voting power is capped below 9.9%; Rumble has also acquired Tether's Northern Data and counts Tether as its most important advertising client, committed to $50 million annually for two years. Tether and Bitfinex are controlling shareholders of Twenty One Capital (XXI), with rights to nominate four of seven directors, including Tether CEO Paolo Ardoino; the company trades at about 60% of its bitcoin value. In gold, Tether holds about 27% of Elemental Royalty (ELE), 12.6% of Gold.com (GOLD), 12.6% of Versamet Royalties (VMET), 14.6% of Metalla (MTA), 14.4% of Gold Royalty (GROY), 12% of Aurelion (AURE), and about 8% of Antalpha (ANTA). It also holds 19.7% of Bitdeer (BTDR) and a stake in Adecoagro (AGRO), plus shares in Satellogic (SATL), 9.9% of SDEV (down about 99%), and a stake in VCI Global (VCIG) via Oobit. Powerlaw Corp (PWRL) claims roughly $20 million in Tether equity. The exposure is concentrated in a few large holdings—Twenty One, Rumble, and Adecoagro—showing Tether's growing role as a multi-sector financial institution that also places executives in key positions.

marsbit1h ago

4.8 Billion Dollars: Tether's Hand Has Reached Into 13 U.S. Public Companies

marsbit1h ago

Trading

Spot

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of SOL (SOL) are presented below.

活动图片