OpenAI Developer Community
community.openai.com.web.brid.gy
OpenAI Developer Community
@community.openai.com.web.brid.gy
Connect with other developers building with the OpenAI API Platform.

[bridged from https://community.openai.com/ on the web: https://fed.brid.gy/web/community.openai.com ]
Why and how is it even logical that o1 is 10x+ more expensive than o3, when o3 is supposed to be better?
You’re assuming something that has never really been true for OpenAI’s API pricing: > “Newer + stronger model ⇒ must be more expensive.” Historically that’s just wrong. Take gpt-4-32k as a concrete example. Its official pricing was: * $60 / 1M input tokens * $120 / 1M output tokens That’s dramatically more expensive than both o1 ($15 / $60) and o3 ($2 / $8). And yet gpt-4-32k is strictly weaker than the current frontier models. So if you apply your logic consistently, the old, weaker gpt-4-32k being 3–10x the price of today’s stronger models would make even less sense. But that’s exactly how pricing has looked in the past: newer models tend to become both better and cheaper over time. The reason is simple: * Pricing is about product tier and compute cost, not “publishing order.” * o1 is a specialized, slower, more compute-heavy deliberate reasoning product. * o3 is optimized to be the default workhorse: fast, cheap, good-enough reasoning for most use cases. If you don’t need the deliberate behavior of o1, just use o3 and enjoy the fact that in 2025 you’re paying a fraction of what people paid for gpt-4-32k, for far better performance. The “why isn’t the newer model more expensive?” complaint ignores the entire history of API pricing.
community.openai.com
December 11, 2025 at 4:01 AM
Degraded Performance on gpt-4.1-mini ft
Important to note: * JSON mode and JSON schema mode highly encourages “pretty” multi-line output. This was a poor decision by OpenAI, has no “off” switch in a “strict” enforcement, and the tabs and linefeeds trained in doing so is a direct cause of this persistent looping symptom, seen since gpt-4-turbo. * OpenAI should not mess with live models or AI inference. This shows continued disrepect for developers. * The fault in model inference is severe here. You’ve shown the AI continuing after a closed JSON, and also not even writing a key’s value but instead emitting linefeeds within a JSON. * Sampling fixes you can apply to workaround this fault in inference, such as stop sequences and logit_bias are also missing in Responses. Rejecting that poor endpoint is wise. Let’s imagine your application prompted instead of trained (and training should have a good system prompt anyway to depart from typical AI behavior). Your `json_schema` object upgraded to “strict”, with formatting undamaged by the forum (chat playground format): { "name": "angels_extraction", "strict": true, "schema": { "type": "object", "properties": { "investors": { "type": "array", "items": { "type": "object", "properties": { "angel_investor_name": { "type": "string", "description": "The name of the angel investor that is investing" }, "lead_investor": { "type": "boolean", "description": "True if the angel investor was a lead investor in the deal. False otherwise" }, "financing_id": { "type": "string", "description": "The financing id that matches the angel investor to the financing" } }, "required": [ "angel_investor_name", "lead_investor", "financing_id" ], "additionalProperties": false } } }, "required": [ "investors" ], "additionalProperties": false } } Output: # Mitigations: * retraining on the JSON format that is enforced by a schema * ensure you use newlines and tabs exactly as the responses you get by strict structured output * retraining on the JSON when there are no investors * reinforcement learning on empty JSON cases may help to produce and close the empty array correctly, as well as avoiding hallucination and fabrication * an anyOf subschema when there is no output * you can receive a different trigger with a simple key/string when the AI wants to report entity extraction failure * stop sequences * you can halt on an immediately closed investors array, and take the “stop” reason as no results * you can halt on excessive linefeeds or even two in a row, indicating the damage. You can strip() to see if there is a useful JSON string within, or close it. * migrate * discover a place where your applications are respected, or that you can use open-weight models under your control. Python for a 'strict' response_format field (click for more details)
community.openai.com
December 11, 2025 at 2:01 AM
Back to the future with people as the catalyst
Subject: Gemini 3 is about fear of losing out. When fear determines the future, the future returns to the past. It kills dreams. Open AI and ChatGpt is about more. Lead by example a clear Path balancing AI Innovation with Human Creativity. Core Proposal: Provide a shift in AI development that combines geometric reasoning with a human-centric interface to help assist the billions of small businesses across the world and individuals who are looking for THEIR ideas to spark a creative revolution in their world. I am an attorney and have been working in healthcare for many years on what many consider the lower end - Community Health Centers a/k/a Federally Qualified Health Centers. Physicians and attorneys, for example, will not give up their intuitive human skills easily in favor of AI taking their place. Even if not the goal, it is the perception. OpenAI and now others have created exotic tools that few honestly understand. The approach you began with years ago empowers users to explore, innovate, and receive confidence-based answers, providing both definitive responses and open-ended exploration. Remove yourself for a moment from the “mouse wheel.” Follow the needs of the people who inspire exploration - individual scientific and philanthropic imagination. IT has been said many ways but it always comes down to the same thing that James Carville has consistently publicly stated: ITts about the economy, stupid. I always has been and always will be – until it’s not. Why This Matters: This is a strategic alignment with OpenAI’s original vision: to lead by thinking differently. It is much more than a technical re-alignment. Offer a path where AI is a partner in human growth, not just a provider of answers. This balanced model ensures that OpenAI stays ahead by embracing both innovation and the human imagination. How? Proposal: A Distinct, Simpler Paradigm for AI Reasoning with a few non-traditional “twists” 1. Geometric Confidence Fields Represent knowledge as geometric entities and measure confidence through spatial tension. 2. Inference as a Core Module Use a distinct inference engine to maintain transparency and modular reasoning. 3. Socratic User Interaction Engage users through questions, keeping human judgment central. 4. Secondary Spiral Integration Re-evaluate and integrate outliers over time through a spiral process, ensuring flexibility. 5. GPU Efficiency via Forward-Looking Application Apply a subset of this approach to GPU control, optimizing how computational resources are used. This can significantly reduce the need for extensive hardware by making existing resources more effective. Importantly, this is a forward-looking suggestion: rather than retrofitting old systems, implement these principles in future developments while letting existing systems continue as they are. A Reminder: Real people need help defining their own aspirations and ideas. Do not forget why OpenAI started. The future depends on thinking outside the usual constraints. By doing so, OpenAI can lead a future where AI and humanity evolve together. Turn potential into lasting impact. Return to your roots – Balance AI Innovation and Human Creativity. Gemini will become little more than a distant dream of those who know no better. Offer a path where AI is a partner in human growth, not just a provider of answers. This balanced model ensures that OpenAI stays ahead by embracing both innovation and the human imagination. Don’t lose the race. Lead it with a human-AI collaboration. It’s about the people. The present road is nothing more than a necessary evil to demonstrate the effects of a past that does not serve humanity for the future. This is not sentiment. It is reality. AI as crowdsourced people is the way forward. Ignoring the concept is a loser because when the dream begins to fade, investor interest will follow. the same course. In the end, “no money, no mission.”
community.openai.com
December 11, 2025 at 12:00 AM
I'm struggling to get the Ballad TTS voice to work
Here is an example: Input: > 8 Leading Restaurants In Los Angeles > > 1. Enaka > > > Enaka, located in Palms, represents one of the most acclaimed expressions of modern kaiseki outside Japan. Under the direction of chef Niki Nakayama, the restaurant offers a meticulously choreographed tasting menu that marries classic Japanese technique with Californian seasonality, often highlighting local seafood, immaculate vegetables, and thoughtful touches such as house-made noodles or inventive sashimi preparations. Service moves at a discreet, attentive, almost meditative pace, and the intimate dining room reinforces the exceptional nature of the experience. The balance among flavors, textures, and the pacing of the meal, combined with an influential wine and sake program, has made Enaka a benchmark of fine dining in Los Angeles and a standard-bearer for Japanese cuisine in the United States. > > 2. Providence > > > Providence, in Hollywood, stands as one of the city’s most enduring temples of seafood cuisine. Chef Michael Chimarusti builds his menu around sustainable fish and seafood, often line-caught, prepared with a refined, French-inspired sensibility and an emphasis on the purity of flavor. The multi-course tasting menus range from delicate crudos and caviar to deeply flavored hot dishes and elaborate desserts, all supported by a serious, international wine list and service that is professional yet warm. The serene, understated dining room keeps the focus squarely on the plate, and Providence’s long history of critical acclaim has cemented its status as a flagship of high-end dining in the city. > > 3. Bestia > > > Bestia, located in the Arts District, helped define the current generation of Los Angeles restaurants with its chic, bustling, industrial setting and its bold, rustic Italian cooking. Chef Ori Menashe’s cooking showcases house-made charcuterie, wood-fired pizzas, fresh pastas with assertive sauces, and generous nose-to-tail meat dishes, all marked by confident seasoning and the character of live fire. The dining room’s effervescence, the open kitchen, and a bar that leans heavily on Italian aperitifs and creative cocktails give it the feel of a modern, lively trattoria on a downtown scale. At Bestia, the combination of serious culinary ambition with an informal, high-energy atmosphere has made it a touchstone for contemporary Italian restaurants across the region. > > 4. Bavel > > > Bavel, Bestia’s Middle Eastern sibling in the Arts District, channels the culinary traditions of Israel, North Africa, and the broader Levant through a polished, produce-focused lens. The menu highlights vibrant spreads and dips, laffa and pita breads coming straight out of the wood-fired oven, intricately spiced vegetable dishes, and large, impressive cuts of meat and seafood, often perfumed with smoke, herbs, and preserved citrus. The bright, plant-filled space is both relaxed and sophisticated, and the drink list offers a bold selection of wines, many from Mediterranean regions, along with cocktails accented with spices. Bavel’s ability to present complex, layered flavors in a modern setting has made it one of Los Angeles’ most talked-about restaurants since it opened. > > 5. Republique > > > Republique, in the heart of Mid-Wilshire, occupies a historic, cathedral-like building originally designed for Charlie Chaplin, and the restaurant makes full use of this spectacular setting. By day, it functions as a bustling bakery and café, known for its pastries, breads, and exceptional casual dining options. In the evening, it transforms into a brasserie-style dining room, where chef Walter Manske’s menu draws on French technique filtered through the California pantry, highlighting impeccable produce, house-made charcuterie, and an evolving selection of meat and seafood dishes that blend comfort with precision. The expansive wine list and the pastry program, led by chef Margarita Manske, add even more depth, making Republique both a neighborhood fixture and a destination restaurant. > > 6. Anajak Thai > > > Anajak Thai, in Sherman Oaks, has evolved from a long-running family restaurant into one of the most influential Thai kitchens in the country. Under chef Justin Pichittrongsi, the cooking moves fluidly between deeply rooted Thai flavors and playful, California-inspired creativity, with a menu that can juxtapose classic curries and grilled skewers with inventive specials and dishes designed to pair with natural wines. The wine program itself is remarkably forward-thinking, with a particular emphasis on low-intervention producers that pair well with spice and fermented notes. Anajak’s intimacy, combined with its blend of tradition and experimentation, has earned it strong critical praise and placed it at the center of conversations about contemporary Thai-American cuisine. > > 7. Pidja Palace > > > Pidja Palace, in the Silver Lake neighborhood, embodies a characteristically Los Angeles fusion of cultures with an Indian-inspired sports bar and a tightly focused menu. The kitchen reinvents bar and comfort food classics through a South Asian lens, offering items such as tandoori-accented chicken wings, creative pastas with Indian flavor profiles, and the namesake pizzas topped with chutney, spiced meats, and vibrant vegetables. Sports broadcasts, clean, minimalist design, and a concise yet thoughtful drink list give it the relaxed feel of a neighborhood hangout, while the food reflects genuine culinary seriousness. This combination of playful concepts, precise execution, and accessible pricing has made Pidja Palace an emerging favorite in the city’s dining landscape. > > 8. Jelaghetza > > > Jelaghetza, in Koreatown, is a foundational institution of Oaxacan cuisine in Los Angeles and a reference point for regional Mexican cooking in the United States. The menu centers on complex moles with slowly developed flavors, generously topped playudas loaded with meat and Oaxacan cheese, rich stews, and traditional drinks such as mezcal and tejate. Live music and a warm, family-friendly atmosphere reinforce the restaurant’s role as a community gathering place and showcase for Oaxacan culture. Recognition from the national press and awards organizations has only amplified its reputation, but Jelaghetza’s core identity remains rooted in generous hospitality and a faithful representation of its culinary heritage. **Code** : {“model”: “gpt-4o-mini-tts”, “input”: Input, “voice”: “Ballad”, “instructions”: “Speak in a clear, deliberate, and professional tone.”} **Note:** You should really check out https://www.openai.fm/ There is a code switch on the top-right of the screen.
community.openai.com
December 11, 2025 at 12:00 AM
Degraded Performance on gpt-4.1-mini ft
from playground too
community.openai.com
December 11, 2025 at 12:01 AM